High Plains Aquifer Calibration Monitoring Well Program: Fourth Year Progress Report

Kansas Geological Survey
High Plains Aquifer Calibration Monitoring Well Program:
Fourth Year Progress Report
R. Stotler, J.J. Butler, Jr., R.W. Buddemeier, G.C. Bohling,
S. Comba, W. Jin, E. Reboulet, D.O. Whittemore, and B.B. Wilson
with contributions by J. Munson and D. Means (KDA-DWR)
Kansas Geological Survey Open-file Report No. 2011-4
March 2011
The University of Kansas, Lawrence, KS 66047 (785) 864-3965; www.kgs.ku.edu
KANSAS GEOLOGICAL SURVEY
OPEN-FILE REPORT
>>>>>>>>>> NOT FOR RESALE <<<<<<<<<<
Disclaimer
The Kansas Geological Survey made a conscientious effort to ensure the accuracy of this report.
However, the Kansas Geological Survey does not guarantee this document to be completely free
from errors or inaccuracies and disclaims any responsibility or liability for interpretations based
on data used in the production of this document or decisions based thereon. This report is
intended to make results of research available at the earliest possible date, but is not intended to
constitute formal publication.
Acknowledgments
We are grateful for the support, assistance, and cooperation of the staff of the Kansas Water
Office, the Kansas Department of Agriculture, Division of Water Resources, the managers and
staff of Groundwater Management Districts 1, 3, and 4, and especially for the cooperation of
Jarvis Garetson (the Garetson Brothers), KBUF, Inc., and Steve and Marilyn Friesen in making
their properties available for installation of the wells. Mark Schoneweis assisted with graphics.
ShyAnne Mailen assisted with final formatting. Susan Stover and Diane Coe of the Kansas
Water Office provided instructive comments on drafts of this report. This project is funded by
the State of Kansas Water Plan Fund.
i
Table of Contents
1.
Introduction and Background ..................................................................................1
2.
Experimental Design ................................................................................................2
3.
Site Descriptions ......................................................................................................2
4.
Overview of Index Well Sites and Monitoring Data ...............................................4
4.1.
Haskell County................................................................................................................ 5
4.1.1.
Hydrograph and General Observations ................................................................... 6
4.1.2.
Measurement Comparisons ..................................................................................... 7
4.2.
Scott County................................................................................................................... 9
4.2.1.
Hydrograph and General Observations ................................................................... 9
4.2.2.
Measurement Comparisons ................................................................................... 11
4.3.
Thomas County ............................................................................................................. 13
4.3.1.
Hydrograph and General Observations ................................................................. 14
4.3.2.
Measurement Comparisons ................................................................................... 15
5.
Water-level Correction...........................................................................................16
5.1.
Water-level Responses to Change in Barometric Pressure ........................................... 16
6.
Thomas County Expansion Project ........................................................................18
7.
Recovery Water-level Estimation Refinement ......................................................23
7.1.
Index Well Horner Recovery Plots ............................................................................... 25
7.2. Seeking Equilibrium: Evaluation of Two-stage Recovery and Recovered Waterlevel Estimation, Thomas County Index Well ...............................................................30
7.3.
Summary: Recovery Water-level Estimation ............................................................... 35
8.
Well Hydrographs and Use of Wells of Opportunity ............................................36
9.
Temporal and Regional Trends: Water Levels and Water Use .............................37
10.
Spin-offs and Related Research .............................................................................41
10.1. Rawlins and Stevens Counties ...................................................................................... 41
10.1.1. Rawlins County..................................................................................................... 41
10.1.2. Stevens County ..................................................................................................... 45
10.2. Haskell County NSF Project ......................................................................................... 46
10.3. Department of Energy Grant – NMR Investigations of Index Wells ........................... 47
11.
Summary and Conclusions ....................................................................................47
12.
References ..............................................................................................................49
Appendix A: Haskell Co. Recovery Plots......................................................................................50
Appendix B: Thomas Co. Recovery Plots .....................................................................................82
Appendix C: Using the KGS Barometric Pressure Correction Spreadsheet and Related
Software .................................................................................................................91
Appendix D: KGS Four Township Thomas Co. Region Water Budget Study ...........................103
Introduction ............................................................................................................................. 103
Background ............................................................................................................................. 104
Data Used ................................................................................................................................ 106
Budget Components – Description and Assessment: ............................................................. 106
Water Elevations and Changes ........................................................................................... 106
Water Extracted .................................................................................................................. 108
Precipitation Data................................................................................................................ 109
Groundwater Flow .............................................................................................................. 111
Groundwater Flow Velocities ................................................................................................. 114
ii
Analysis and Discussion ......................................................................................................... 116
Reported Use ....................................................................................................................... 116
Recharge and Other Factors ................................................................................................ 123
Water Budget ...................................................................................................................... 125
Conservation and Management Implications...................................................................... 128
Reducing Uncertainties: Data Needs .................................................................................. 129
Summary and Conclusions ..................................................................................................... 131
Appendix D-1 – v2 (Geographic Sample) Change vs. Use Regressions ................................ 133
Appendix D 2 .......................................................................................................................... 135
Appendix D-3......................................................................................................................... 137
Appendix E: New Insights From Well Responses to Fluctuations In Barometric Pressure ........147
ABSTRACT ............................................................................................................................ 148
INTRODUCTION .................................................................................................................. 148
Field Site Overview ................................................................................................................ 151
Methodology ........................................................................................................................... 152
Application .............................................................................................................................. 157
Discussions and Conclusions .................................................................................................. 160
REFERENCES ....................................................................................................................... 163
APPENDIX E1 ....................................................................................................................... 166
iii
LIST OF FIGURES
Figure 1: The Kansas portion of the High Plains aquifer .............................................................4
Figure 2: Haskell County site .......................................................................................................5
Figure 3: Haskell County index well hydrograph. ........................................................................6
Figure 4: Scott County site ...........................................................................................................9
Figure 5: Scott County index well hydrograph. ..........................................................................10
Figure 6: Thomas County site .....................................................................................................13
Figure 7: Thomas County index well hydrograph ......................................................................14
Figure 8: Thomas index well hydrograph, with corrected water levels ......................................18
Figure 9: Hydrograph comparison from the Thomas Expansion Well program .......................20
Figure 10: Groundwater elevation contours near the Thomas index well ..................................21
Figure 11: Barometric pressure response function for Thomas Co. expansion wells
TH7, TH9, and TH10. ...............................................................................................22
Figure 12: Well hydrograph, barometric pressure, and water-level corrected for
barometric response from Thomas Co. expansion well TH7. ..................................23
Figure 13: HS 20 hydrograph. ....................................................................................................24
Figure 14: HS 20 Horner recovery plot from August 30, 2007 to March 20, 2008, with
early recovery and late recovery estimates ...............................................................25
Figure 15: Haskell index well recovery for all three complete recovery seasons (200708, 2008-09, 2009-10) and the start of the 2010-11 recovery ..................................26
Figure 16: Scott index well recovery for all three complete recovery seasons (2007-08,
2008-09, 2009-10) and the start of the 2010-11 recovery. .......................................28
Figure 17: Thomas index well recovery for all three complete recovery seasons (200708, 2008-09, 2009-10) and the start of the 2010-11 recovery. .................................29
Figure 18: Hydrograph of the 2009 Thomas index well pumping season, and the
subsequent 2009-10 recovery ...................................................................................30
Figure 19: 2009-10 Thomas index well recovery estimation plot, displaying the
differences between recovery estimates considering the full pumping season
(SP, tp=63d) and the average pumping time of a single well (SW, tp=5d) ...............31
Figure 20: 2009-10 Thomas index well recovery estimation plot, displaying the
differences between recovery estimates considering data from differing late
time periods ...............................................................................................................33
Figure 21: Hydrographs of wells 23289, obs23289, and obs28290 in 3S-36W-10,
Rawlins Co. ...............................................................................................................42
Figure 22: Measured water levels, barometric pressures, and corrected water levels for
wells 23289, obs23289, and obs28290 in 3S-36W-10, Rawlins Co.........................43
Figure 23: Barometric response functions for wells 23289, obs23289, and obs28290, in
3S-36W-10, Rawlins Co. ..........................................................................................44
Figure 24: Hydrographs for wells 42421, 42423, 42453, 40578, obs40578, and 44593 in
Stevens County .........................................................................................................46
iv
Appendix A.
Figure A - 1: Horner recovery estimation HS1, (a) 2007-08; and 2008-09: (b) entire
recovery period, (c) only the recovery after the final pumping event. ...............51
Figure A - 2: Horner recovery estimation HS2, 2007-08; (A) entire recovery period, (B)
only the recovery after the final pumping event. ................................................52
Figure A - 3: Horner recovery estimation HS2, 2008-09; (A) entire recovery period, (B)
only the recovery after the final pumping event. ................................................53
Figure A - 4: Horner recovery estimation HS2, 2009-10. ..........................................................54
Figure A - 5: Horner recovery estimation HS3, (a) 2007-08, (b) 2008-09, (c) 2009-10. ..........55
Figure A - 6: Horner recovery estimation HS4, 2008-09. ..........................................................56
Figure A - 7: Horner recovery estimation HS4, 2009-10. ..........................................................56
Figure A - 8: Horner recovery estimation HS5, 2007-08. ..........................................................57
Figure A - 9: Horner recovery estimation HS5, 2008-09; (A) entire recovery period, (B)
only the recovery after the final pumping event. ................................................58
Figure A - 10: Horner recovery estimation HS5, 2009-10. ........................................................59
Figure A - 11: Horner recovery estimation HS6, 2007-08; (A) entire recovery period, (B)
only the recovery after the final pumping event. ................................................60
Figure A - 12: Horner recovery estimation HS6, 2008-09. ........................................................61
Figure A - 13: Horner recovery estimation HS6, 2009-10. ........................................................61
Figure A - 14: Horner recovery estimation HS7, 2007-08. ........................................................62
Figure A - 15: Horner recovery estimation HS7, 2008-09; (A) entire recovery period, (B)
only the recovery after the final pumping event. ................................................63
Figure A - 16: Horner recovery estimation HS7, 2009-10. ........................................................64
Figure A - 17: Horner recovery estimation HS8, 2007-08; (A) entire recovery period, (B)
only the recovery after the final pumping event. ................................................65
Figure A - 18: Horner recovery estimation HS8, 2008-09. ........................................................66
Figure A - 19: Horner recovery estimation HS8, 2009-10. ........................................................66
Figure A - 20: Horner recovery estimation HS9, (a) 2007-08, (b) 2008-09, (c) 2009-10. .........67
Figure A - 21: Horner recovery estimation HS11, 2008-09; (a) entire recovery period, (b)
only the recovery after the final pumping event, and (c) 2009-10......................68
Figure A - 22: Horner recovery estimation HS12, 2007-08. ......................................................69
Figure A - 23: Horner recovery estimation HS15, 2007-08. .....................................................70
Figure A - 24: Horner recovery estimation HS15, 2009-10. ......................................................70
Figure A - 25: Horner recovery estimation HS17, 2007-08; (A) entire recovery period,
(B) only the recovery after the final pumping event. ..........................................71
Figure A - 26: Horner recovery estimation HS18, 2007-08; (A) entire recovery period,
(B) only the recovery after the final pumping event. ..........................................72
Figure A - 27: Horner recovery estimation HS18, 2008-09; (A) entire recovery period,
(B) only the recovery after the final pumping event. ..........................................73
Figure A - 28: Horner recovery estimation HS18, 2009-10. ......................................................74
Figure A - 29: Horner recovery estimation HS20, 2007-08; (A) entire recovery period,
(B) only the recovery after the final pumping event. ..........................................75
Figure A - 30: Horner recovery estimation HS20, 2009-10; (A) entire recovery period,
(B) only the recovery after the final pumping event. ..........................................76
Figure A - 31: Horner recovery estimation HS21, 2008-09. ......................................................77
v
Figure A - 32: Horner recovery estimation HS21, 2009-10. ......................................................77
Figure A - 33: Horner recovery estimation HS28, 2008-09. ......................................................78
Figure A - 34: Horner recovery estimation HS29, 2008-09. ......................................................79
Figure A - 35: Horner recovery estimation HS30, 2007-08; (A) entire recovery period,
(B) only the recovery after the final pumping event. ..........................................80
Figure A - 36: Horner recovery estimation HS31, (a) 2007-08, (b) 2008-09, and (c) 200910.........................................................................................................................81
Appendix B.
Figure B - 1: Thomas Co. index well hydrograph, barometric pressure, and corrected
water level, 2007-08 recovery.............................................................................83
Figure B - 2: Horner recovery estimations, Thomas Co. index well, 2007-08 recovery
season. .................................................................................................................83
Figure B - 3: Horner recovery estimations, Thomas Co. index well, following pumping
periods #1 and #2 in 2008. ..................................................................................84
Figure B - 4: Thomas Co. index well hydrograph, barometric pressure, and corrected
water level, 2008-09 recovery.............................................................................85
Figure B - 5: Horner recovery estimations, Thomas Co. index well, 2008-09 recovery
season. .................................................................................................................85
Figure B - 6: Horner recovery estimations, Thomas Co. index well, following pumping
periods #1, #2, and #3 in 2009. ...........................................................................86
Figure B - 7: Thomas Co. index well hydrograph, barometric pressure and corrected
water level, 2009-10 recovery.............................................................................87
Figure B - 8: Horner recovery estimations, Thomas Co. index well, 2009-10 recovery
season. .................................................................................................................87
Figure B - 9: Thomas Co. index well hydrograph and corrected water level, 2010-11
recovery...............................................................................................................88
Figure B - 10: Horner recovery estimations, Thomas Co. index well, 2010-11 recovery
season. .................................................................................................................88
Figure B - 11: 2009-10 hydrograph (A) and recovery from Thomas Co. well TH3.
Recovery is plotted as semi-log recovery (B) and as Horner recovery (C)
and (D) ................................................................................................................89
Figure B - 12: 2009-10 hydrograph (A) and recovery from Thomas Co. well TH10.
Recovery is plotted as semi-log recovery (B) and as Horner recovery (C)
and (D) ................................................................................................................90
Appendix D:
Figure D - 1: Townships 9-32, 9-33, 9-34 and 10-33 in Thomas County, with surrounding
area ....................................................................................................................104
Figure D - 2: Terms in the water budget of a region ...............................................................105
Figure D - 3: TIN boundaries used to calculate water levels that fall between the
measuring points. ..............................................................................................107
vi
Figure D - 4: Reported water use for the four townships ........................................................109
Figure D - 5: Total annual precipitation values for the years 1996-2004 for the three
weather stations in or near the area of interest ..................................................110
Figure D - 6: Rainfall during the months associated with the growing season .......................111
Figure D - 7: Example of west to east cross section showing elevations of land surface,
predevelopment water table, average 1996-2005 water table, and bedrock
surface ...............................................................................................................112
Figure D - 8: Groundwater elevation contours ........................................................................113
Figure D - 9: Schematic illustration of the estimated groundwater flow relations among
the four Thomas County townships ..................................................................114
Figure D - 10: Well hydrographs for the monitoring wells characterizing each township......118
Figure D - 11: Annual change converted to estimated AF/Twp, vs. reported use for each
township. ...........................................................................................................122
Figure D - 12: Average 2-mile use density, AF/section ..........................................................123
Figure D - 13: Aspects of groundwater recharge .....................................................................124
Figure D - 14: Monitoring water-level change in undeveloped areas .....................................131
vii
LIST OF TABLES
Table 1: Characteristics of the index well sites. .......................................................................... 3
Table 2: General characteristics of the Haskell Co. index well hydrograph and local
water-use data. ........................................................................................................... 7
Table 3. Annual water-level measurement comparison with transducer measurements,
Haskell Co. ................................................................................................................. 8
Table 4: General characteristics of the Scott index well hydrograph and local water-use
data. .......................................................................................................................... 11
Table 5: Annual water-level measurement comparison with transducer measurements,
Scott Co.................................................................................................................... 12
Table 6: General characteristics of the Thomas index well hydrograph and local wateruse data..................................................................................................................... 15
Table 7: Annual water-level measurement comparison with transducer measurements,
Thomas Co. .............................................................................................................. 16
Table 8: Installation date and other notes for Thomas Co. expansion wells. ............................ 19
Table 9: Average water-level elevation (barometric effects not corrected), HS 20. ................. 25
Table 10: Comparison of reference water levels used in the Horner plot recovery analysis
for Haskell County ................................................................................................... 27
Table 11: Comparison of observed and predicted recovery estimations for the 2009-10
recovery of the Thomas index well. ......................................................................... 32
Table 12: Summary of recovered water-level estimates and observed values, Thomas
index well ................................................................................................................. 34
Table 13: Water use and recovered water levels at the Three Index Well Sites. ....................... 38
Table 14: Water use and provisional recovered water levels in wells near the Haskell
County index well. ................................................................................................... 39
Table 15: Statistical summary of water use and recovered water level from Table 14 for
wells in the vicinity of the Haskell index well. ........................................................ 41
Appendix D
Table D - 1: Acre-feet per year of reported use for the four townships...................................108
Table D - 2: Precipitation measurements in and near the Thomas County area of interest .....110
Table D - 3: Summary of net groundwater fluxes ...................................................................113
Table D - 4: Groundwater flow velocity estimates ...................................................................115
Table D - 5: Summary of average budget term estimates for the period 1996-2004...............127
Table D - 6: Uncertainties in net water balance for the TIN v2case with two assumptions
about the uncertainty of the input data. .............................................................128
viii
1. Introduction and Background
The calibration monitoring (index) well program is a pilot study to develop improved approaches
for measuring and interpreting hydrologic responses at the local (section to township) scale in the
Ogallala-High Plains aquifer (henceforth, High Plains aquifer). The study is supported by the
Kansas Water Office (KWO) with Water Plan funding as a result of KWO’s interest in and
responsibility for long-term planning of groundwater resources in western Kansas. The Kansas
Department of Agriculture, Division of Water Resources (KDA-DWR), is providing assistance,
in terms of personnel and equipment, as are Groundwater Management Districts (GMDs) 1, 3,
and 4.
A major focus of the program is the development of criteria or methods to evaluate the
effectiveness of management strategies at the sub-unit (e.g., township) scale. Changes in water
level – or the rate at which the water level is changing – are considered the most direct and
unequivocal measure of the impact of management strategies. Because of the economic, social,
and environmental importance of water in western Kansas, the effects of any modifications in
patterns of water use need to be evaluated promptly and accurately. The project has focused on
identifying and reducing the uncertainties and inaccuracies in estimates of year-to-year changes
in water level, so that the impacts of management decisions can be assessed as rapidly as
possible. The approach outlined by this study aims to provide more accurate and timely
information at the sub-unit scale than is provided by the annual water-level measurement
program. Furthermore, this study provides data that are valuable for the interpretation (or
calibration) of the water-level change estimates from the annual measurement program.
At the end of year four of the study, monitoring data from three full recovery and pumping
seasons and the start of a fourth recovery season have been obtained. With increasing data, the
index well program has demonstrated that (1) the annual water-level measurement network (even
with additional semi-annual observations) does not currently produce an adequate dataset to
evaluate how management decisions affect water-level changes in the short term (fewer than five
years); (2) because of uncertainties in both the effects of barometric pressure changes and the
degree of well recovery at the time of the annual water-level measurement program, the data
from the index wells provide the context needed for the interpretation of the results of the annual
measurement program; (3) additional measurements at nearby [local (~township) scale] wells are
needed through most of a recovery season to establish the representative area (areal reach) of an
index well; (4) with a complete recovery record, it appears possible to extrapolate to fully
recovered water levels; (5) local hydrogeologic variations and well construction need to be
assessed and considered in the interpretation of well hydrographs; these factors may complicate
use of wells of opportunity as index wells; and (6) water-level data collected using a pressure
transducer and data logger provide a near-continuous water-level record that can help in the
estimation of changes in the amount of extractable water and in assessing the uncertainty in those
estimates.
This report will provide (a) an update of the hydrographs for the three index wells; (b) a detailed
look at methods to estimate the elevation to which the water level would rise at full recovery in
each of the index wells; (c) comparisons of the annual water-level changes measured at the index
1
wells with those from nearby wells to assess the representative area sampled by the index wells;
(d) interpretation of hydrographs from the index wells and the wells in the expanded monitoring
areas in the vicinity of two of the index wells; and (e) an overview of the KGS barometric
correction spreadsheet program, which calculates the barometric response function for a given
well and corrects the measured water levels for the impact of barometric pressure changes.
2. Experimental Design
The foundation of the experimental component of the project consists of three transducerequipped wells, designed and sited to function as local monitoring wells, installed in late summer
2007. There is one well in each of the three western GMDs, with locations deliberately chosen
to represent different water use and hydrogeologic conditions, and to take advantage of related
past or current studies (Figure 1). The original experimental design envisioned use of the index
wells to anchor and calibrate the manual measurements of annual program wells in the area near
an index well, thus providing more consistency and confidence in the calculation of the watertable surface and its changes in that general vicinity. However, the findings discussed in KGSOFR 2010-03 (Buddemeier et al., 2010) led to the realization that more extensive measurements
and calibration were necessary to develop a suitable measurement protocol. To achieve this, the
project has been expanded to include “wells of opportunity” in the vicinity of two of the index
wells:
1. The Haskell site, with numerous other wells instrumented by KDA-DWR, provides an
opportunity for more extensive comparisons over a relatively short distance. However,
the fact that the producing wells at the Haskell site may draw on and measure either or
both of two separate aquifer units makes it more complicated than the commonly adopted
view of the High Plains aquifer as a single unconfined aquifer.
2. The Thomas site, for which the commonly adopted view of the High Plains aquifer as a
single unconfined aquifer appears appropriate, has been expanded to complement the
comparisons at the Haskell site. With the collaboration of KDA-DWR and GMD4, six
additional wells (two of which are annual program wells) have been equipped with
transducers.
With increasing data, it has become apparent that these expansions enhance confidence in data
gathered from the index wells and in estimates of the areal reach (representative area) of the
index wells. Once a representative area of an index well has been determined, continued
monitoring of the additional wells is not necessary and may be modified or discontinued.
3. Site Descriptions
Site characteristics are described and discussed in detail in previous publications (Young et al.,
2007, 2008; Buddemeier et al., 2010), so they are only briefly summarized below and in Table 1.
The three sites are located, south to north, in Haskell, Scott, and Thomas counties.
2
The Haskell County site represents the most complex set of conditions. It is located over a
relatively steeply sloping portion of the bedrock surface underlying the High Plains aquifer, and
along a gradient in both water use and water availability. Although the saturated thickness is
large, the thickness of intervals that readily yield water to wells is much less. Well yields have
deteriorated as water levels have continued to decline and an impairment complaint (since
withdrawn) was filed before the commencement of this study. It appears that a two-zone aquifer
system exists in the vicinity of the site: an unconfined upper aquifer zone and a thin, but
productive, confined aquifer zone on top of bedrock, with a thick clay layer separating the two.
The project well was installed to sample only the lower confined aquifer zone near the site of the
impairment complaint; KDA-DWR has installed transducers in a number of nearby wells in both
aquifer zones and these wells are also utilized by this project. The Haskell County site is in an
area of greater saturated thickness than the other sites, but with greater lateral variation in aquifer
characteristics and a more rapid rate of water-level decline.
The Scott and Thomas sites are both located in areas where the saturated thickness is generally
100 ft or less, with areas of less than 50 ft nearby. Since 50-100 ft of saturated thickness is
required to sustain high-volume irrigation pumping under most aquifer and water-use conditions
(Hecox et al., 2002) and both areas have shown steady declines in water level, these sites are
vulnerable to resource exhaustion. The Scott County site has the only well that directly monitors
the water level in the northern portion of the Scott-Finney depression, which serves as the major
water supply for Scott City. In addition, Scott County has also recently been the location of a
project that uses analyses of drillers’ logs to determine and map the intervals of the aquifer that
readily yield water (Practical Saturated Thickness Plus (PST+) Project). This information is
important for relating aquifer lithology to well response characteristics. The Thomas County site
has been the subject of previous water budget analyses (Appendix D) and is of additional interest
because of 1) the presence of stream channels (the channel of the South Fork of the Solomon
River runs east-west just north of the index well) that may influence recharge, and 2) the
proximity of the site to the edge of the productive portion of the High Plains aquifer. Both the
Scott and Thomas sites are assumed to represent unconfined (water-table or phreatic) aquifer
conditions, whereas the Haskell site represents confined aquifer conditions.
Table 1: Characteristics of the index well sites.
2010
Bedrock
Screened
Site
2010
Saturated
depth
interval (ft
WL
thickness (estimated ft below lsf)
elev.
(ft)
below lsf)
(ft)a
Haskell
2575.6
170.6
433
420-430
Scott
2835.0
91.0
223
215-225
Thomas 2974.6
71.6
284
274-284
a
2009 Water Use (AF)
1-mi
2-mi
5-mi
circle
circle
circle
1935
873
587
2010 annual tape water-level measurements from WIZARD database
(http://www.kgs.ku.edu/Magellan/WaterLevels/index.html)
3
8720
2955
1917
45754
16427
7335
Figure 1: The Kansas portion of the High Plains aquifer, with aquifer and county
boundaries shown. The colored pixels represent one section (1 mi2), coded to show the
degree of groundwater depletion from the beginning of large-scale development to the
average of conditions in 2007-2009. The three green boxes surround the index well study
sites.
4. Overview of Index Well Sites and Monitoring Data
This section provides a brief overview of the hydrographs from all three sites. With over three
and a third years of hourly measurements, our understanding of water-level responses and trends
at all three sites has improved significantly. All three index well hydrographs indicate that,
although pumping occurs sporadically throughout the year, the major drawdown in water levels
occurs during the pumping season in the summer when the aquifer is stressed significantly for an
extended period of time. For this study, the pumping season is defined as the period from the
first sustained drawdown during the growing season (often, but not always, following the
maximum recovered water level) to the first major increase in water level near the end of the
growing season. The recovery season is defined as the time between pumping seasons. Since
water levels increase throughout the recovery period at all three index wells, and full recovery
has not been observed at any of the wells, the difference between water levels measured during
the recovery season from one year to the next only provides a measure of the year-to-year change
in still-recovering water levels. This year-to-year change in recovering water levels is of limited
value for managers because it can be affected by a variety of factors, such as the duration of
4
recovery at the time of the measurement, that are of little significance for assessing aquifer
trends. More importantly, it does not involve the final recovered water level, the elevation to
which the water level would rise if the recovery was not interrupted by the next pumping season.
This final recovered water level, which would provide a reliable basis for managers to assess the
impact of changes in water use, can only be estimated through various extrapolation procedures.
These extrapolation procedures were a major focus of the project this year, and will be discussed
in detail in Section 7 of this report.
4.1.
Haskell County
Figure 2: Haskell County site, showing the index well, adjacent monitoring wells, and
points of diversion within the area of concentrated KDA-DWR studies. Most of the
marked wells are equipped with transducers.
The Haskell County site is the most extensively monitored of the three sites because of its
location within an area of concentrated KDA-DWR monitoring. Figure 2 is an aerial overview
of the Haskell County site at a scale that shows the index well, the additional wells being
monitored by KDA-DWR and used by the index well program, and the water rights within the
area.
5
4.1.1. Hydrograph and General Observations
The complete hydrograph for the Haskell index well is shown in Figure 3 and its general
characteristics are summarized in Table 2. The confined nature of the aquifer zone in which the
index well is screened is illustrated by the greater than 115 ft. change in water level during each
pumping season, despite the absence of high-capacity pumping wells in the immediate vicinity of
the index well (closest pumping well is almost half a mile away). Each year, the minimum
recorded water-level elevation declined from the previous year. The lowest water level observed
by far was in 2010; the minimum 2010 water-level elevation was 7 ft. lower than in 2008 or
2009 and 8.5 ft. lower than in 2007. This lower minimum water level was obtained despite a
shorter pumping season in 2010 than in 2009. Water use within the 2-mile radius surrounding
the index well was highest during 2008, and approximately 1200 ac-ft less during both 2007 and
2009 (2010 data are not yet available). Each year since the 2007-08 recovery season, the index
well has recorded year-to-year declines in the maximum recovered water level between 4 ft. and
5 ft.. Given the much lower water-level minimum recorded in 2010, the expectation is that the
decline in the maximum recovered water level will exceed the decline observed in previous
years.
Hourly Sensor Measurements
Telemetered 2-hr Measurements
E-tape Measurements
Annual Water Level Measurement
2587
2577
2567
2557
2547
2537
2527
2517
2507
2497
2487
2477
2467
2457
2447
1Au
g07
31
-O
ct
-0
7
30
-J
an
-0
8
30
-A
pr
-0
8
31
-J
ul
-0
8
30
-O
ct
-0
8
29
-J
an
-0
9
1M
ay
-0
9
31
-J
ul
-0
9
30
-O
ct
-0
9
30
-J
an
-1
0
1M
ay
-1
0
31
-J
ul
-1
0
30
-O
ct
-1
0
30
-J
an
-1
1
Elevation of Water Level (ft AMSL)
Haskell Co Index Well
27S 31W 36BDC 01
Figure 3: Haskell County index well hydrograph – total data run to 1/11/11. From 11/2/10
to 1/11/11, the provisional 2-hour telemetered data are used; before that period, data are
hourly downloaded measurements.
6
Table 2: General characteristics of the Haskell Co. index well hydrograph and local wateruse data.
Mimimum WaterLevel Elevation
Maximum
Observed
Recovery
Elevation
Apparent
Recovery
Annual Change in
Maximum
Observed
Recovery
Recovery Season
Pumping During
Recovery Season
Length of
Pumping Season
2-mi. Radius
Water Use
Feet
Date
Feet
2007
2462.2
8/23/07
NA
2008
2460.8
8/8/08
2586.1
2009
2460.7
8/16/09
2581.1
2010
2453.8
8/9/10
2577.2
Date
NA
2/28/08
2/9/09
3/5/10
Feet
NA
123.9
120.3
116.5
Feet
NA
NA
-5.0
-3.9
Start
End
Length (#
Days)
NA
NA
8/24/07
2/28/08
8/13/08
2/9/09
8/16/09
3/5/10
NA
188.14
180.46
182.38
# Days
NA
41.5
19.96
5.25
Length
NA
167.33 205.71
(# Days)
Irrigated
6475
7755
6259
Acres
Total Use
8764.01 9931.71 8720.45
(ac-ft)
Use per
Irrigated
1.35
1.28
1.39
Acre (ft)
169.79
NA
NA
NA
4.1.2. Measurement Comparisons
The transducer measurements continue to compare well with the annual steel tape water-level
measurements, indicating that the transducer provides an accurate representation of the
instantaneous water level in the well (Table 3). When the barometric pressure effect is removed
from the transducer measurement (see Section 5.1), the discrepancy between the annual manual
measurements and the transducer measurements increases because the barometric effect has not
been removed from the annual measurements. While the annual measurement program provides
reasonable estimates of the water levels at the Haskell site, estimates of year-to-year changes in
water level based on those measurements will be influenced by barometric pressure effects and
7
incomplete recovery. The accuracy of those estimates is improved somewhat by correcting the
water-level measurements for variations in barometric pressure. This is readily done by applying
the same correction to the manual measurements as calculated for the transducer measurements.
Year-to-year water-level declines based on the annual measurement program were 4.1 ft. and 4.8
ft. between the 07-08 and 08-09 recovery seasons and the 08-09 and 09-10 recovery seasons,
respectively (Table 3). These declines underestimated the water-level change based on the
maximum recovered water level by 0.9 ft. for the first period, and overestimated the change by
0.9 ft. in the second. The primary reason for the difference between the annual water-level
declines calculated from the manual measurement program and the annual declines in the
maximum recovered water level calculated from the index well transducer is that the decline
estimates are based on measurements taken at different points during the recovery season. These
differences in the annual water-decline estimates are the justification for the development of the
extrapolation procedures to estimate the water level at full recovery discussed in Section 7 of this
report.
Table 3. Annual water-level measurementa comparison with transducer measurements,
Haskell Co.
Date
WL elev (ft)
Indicated Annual
Method
b
WL Decline (ft)
1/15/2008 2584.48
NA
Steel tape
c
2584.44
Transducer
1/7/2009
2580.41
4.07 (5.0)
Steel tape
2580.19c
Transducer
2580.10d
Transducer
1/14/2010 2575.63
4.78 (3.9)
Steel tape
c
2575.54
Transducer
2574.51d
Transducer
a
Steel tape measurements are from annual water-level measurement program
(http://hercules.kgs.ku.edu/geohydro/wizard/wizardwelldetail.cfm?usgs_id=373925100395301).
b
Value in () is the decline in the maximum recovered water level measured by the index well
transducer.
c
average of values, not corrected for barometric pressure, 0800-1600
d
average of values corrected for barometric pressure using the KGS barometric pressure
correction program, 0800-1600
8
4.2.
Scott County
Figure 4: Scott County site, showing the index well and adjacent points of diversion.
Figure 4 is an aerial overview of the Scott County site at a scale that shows the index well, the
surrounding network of annual program wells, and the water rights within the area.
4.2.1. Hydrograph and General Observations
The complete hydrograph for the Scott index well is shown in Figure 5 and its general
characteristics summarized in Table 4. The unconfined nature of the aquifer zone in which the
index well is screened is illustrated by the relatively small change and rate of change in water
level during each pumping and recovery season, despite at least two high-capacity pumping
wells within a half mile of the index well. Each year, the minimum recorded water-level
elevation has declined from the previous year, although there was only a small decline between
9
2009 and 2010. The 2010 low was slightly lower than 2009, more than 1 ft. lower than 2008,
and probably more than 2 ft. lower than 2007 (note the index well was drilled during the 2007
recovery so the 2007 low was not recorded). The year-to-year declines in the maximum
recovered water level were 1.3 ft. and 0.4 ft. between the 2007-08 and 2008-09 recovery seasons
and the 2008-09 and 2009-10 recovery seasons, respectively. Similar to the Haskell site, water
use within the 2-mile radius surrounding the index well was highest during 2008, and
approximately 1000 ac-ft less during 2007 and 2009 (2010 data are not yet available).
Scott Co Index Well
18S 33W 01AAA
2837
Hourly Sensor Measurements
Telemetered 2-hr Measurements
E-Tape Measurements
Annual Water Level Measurements
Elevation of Water (ft AMSL)
2836
2835
2834
2833
2832
2831
-0
8
30
-A
pr
-0
8
31
-J
ul
-0
8
30
-O
ct
-0
8
29
-J
an
-0
9
1M
ay
-0
9
31
-J
ul
-0
9
30
-O
ct
-0
9
30
-J
an
-1
0
1M
ay
-1
0
31
-J
ul
-1
0
30
-O
ct
-1
0
30
-J
an
-1
1
an
7
ct
-0
30
-J
31
-O
1-
Au
g-
07
2830
Figure 5: Scott County index well hydrograph – total data run to 1/11/11. From 11/2/10 to
1/11/11, the provisional 2-hour telemetered data are used; before that period, data are
hourly downloaded measurements.
10
Table 4: General characteristics of the Scott index well hydrograph and local water-use
data.
2007
2008
2009
2010
<2833.4
2832.0
2831.2
2830.9
Feet
Minimum Water8/21/07
9/5/08 8/30/09
8/24/10
Level Elevation
Date
and
9/18/10
Maximum
Feet
NA
2835.9
2834.6
2834.2
Observed Recovery
Date
NA
3/4/08
2/17/09
3/2/10
Elevation
Apparent Recovery
Feet
NA
>2.5
2.7
3.0
Apparent WaterLevel Change from
Feet
NA
NA
-1.3
-0.4
Previous Year
Start
NA
<8/21/07 10/11/08
8/30/09
End
NA
3/11/08
4/2/09
4/5/10
Recovery Season
Length (#
NA
>203
204.71
217.79
Days)
Pumping During
# Days
NA
>48.21
13.7
21.04
Recovery Season
Length of Pumping
Length
NA
182.29
150.04
145.67
Season
(# Days)
Irrigated
4132
3950
3923
NA
Acres
Total Use
3175.09
4059.02
2955.48
NA
(ac-ft)
Irrigation
2-mi Radius Water
Use Only 3095.78
4014.33
2955.48
NA
Use
(ac-ft)
Irrigation
Use per
0.75
1.02
0.75
NA
Irrigated
Acre (ft)
4.2.2. Measurement Comparisons
Overall, the annual water-level measurements and the transducer measurements that have not
been corrected for barometric pressure effects showed good agreement in the Scott index well
record (Table 5). In 2008, 2009 and 2010, the discrepancy was 0.00 ft., 0.02 ft. and 0.01 ft.,
respectively.
Year-to-year water-level declines based on the annual well program were 1.1 ft. and 0.7 ft.
between the 07-08 and 08-09 recovery seasons and the 08-09 and 09-10 recovery seasons,
11
respectively (Table 5). These annual program water-level declines underestimated the waterlevel change based on the maximum recovered water level by 0.2 ft. for the first period, and
overestimated the change by 0.5 ft. in the second.
Table 5: Annual water-level measurementa comparison with transducer measurements,
Scott Co.
Method
Indicated
Annual WL
Decline (ft)b
1/7/2008 2835.29
NA
Steel tape
c
2835.29
Transducer
1/6/2009 2834.23
1.06 (1.24)
Steel tape
c
2834.21
1.08
Transducer
2834.95d
Transducer
1/7/2010 2833.49
0.74 (0.28)
Steel tape
2833.48c
0.73
Transducer
2833.55e
1.40
Transducer
a
Steel tape measurements are from annual water-level measurement program
(http://hercules.kgs.ku.edu/geohydro/wizard/wizardwelldetail.cfm?usgs_id=391404101010701)
b
Value in () is the decline in the maximum recovered water level measured by the index well
transducer
c
average of values, not corrected for barometric pressure, 0800-1600
d
back extrapolated (quadratic best fit) from barometrically corrected values, 1/8/2009–2/18/2009
e
average of values, corrected for barometric pressure using the KGS barometric pressure
correction program, 0800-1600
Date
WL elev (ft)
12
4.3.
Thomas County
Figure 6: Thomas County site, showing the index well, nearby wells that have been
equipped with transducers, surrounding annual wells, and points of diversion in the area.
Figure 6 is an aerial overview of the Thomas County site at a scale that shows the index well, the
additional wells in which transducers have been placed, the surrounding network of annual
program wells, and the water rights within the area.
13
4.3.1. Hydrograph and General Observations
The complete hydrograph for the Thomas index well is shown in Figure 7 and its general
characteristics are summarized in Table 6. The unconfined nature of the aquifer zone in which
the index well is screened is illustrated by the relatively small change and rate of change in water
level during each pumping and recovery season, despite 10 or more high-capacity pumping wells
within a mile of the index well. In 2010, the lowest water level was recorded on September 5,
and was nearly the same as in 2009, but 1.3 ft. higher than in 2008. The 2009-2010 recovery
continued until June 22, and constituted the longest period of recovery observed at any of the
index wells. Water levels in 2009-2010 recovered to the highest level recorded to date in the
Thomas index well. Unlike the other two sites, water use within the 2-mile radius surrounding
the index well was similar during 2007 and 2008, and approximately 1000 ac-ft less during 2009
(2010 data are not yet available).
Hourly Sensor Measurements
Telemetered
E-Tape measurements of Depth to Water
Annual Water Level Measurements
Thomas Co Index Well
09S 33W 33BBB
Elevation of Water Level (ft AMSL)
2977
2976
2975
2974
2973
2972
2971
2970
1Au
g07
31
-O
ct
-0
7
30
-J
an
-0
8
30
-A
pr
-0
8
31
-J
ul
-0
8
30
-O
ct
-0
8
29
-J
an
-0
9
1M
ay
-0
9
31
-J
ul
-0
9
30
-O
ct
-0
9
30
-J
an
-1
0
1M
ay
-1
0
31
-J
ul
-1
0
30
-O
ct
-1
0
30
-J
an
-1
1
2969
Figure 7: Thomas County index well hydrograph – total data run, 8/7/07 to 1/11/11. From
11/3/10 to 1/11/11, the provisional 2-hour telemetered data are used; before that period,
data are hourly downloaded measurements.
14
Table 6: General characteristics of the Thomas index well hydrograph and local water-use
data.
Mimimum
Drawdown
Elevation
Maximum
Observed Recovery
Elevation
Apparent Recovery
Apparent WaterLevel Change from
Previous Year
Recovery Season
Pumping During
Recovery Season
Length of Pumping
Season
2-mi Water Use
Date
2007
2970.42
9/2/07
Feet
NA
2975.54
2975.09
2976.20
Date
NA
5/2/08
6/24/09
6/21/10
Feet
NA
5.12
5.38
5.42
Feet
NA
NA
-0.45
+1.11
Start
End
Length
(# Days)
NA
NA
9/8/07
5/12/08
9/8/08
6/24/09
8/27/09
6/21/10
NA
247.21
289.42
298.46
# Days
NA
0?
17.04
2.17
NA
118.46
63.33
77.50
2983
3016
2958
NA
2868.87
2825.21
1917.17
NA
0.96
0.94
0.65
NA
Feet
Length
(# Days)
Irrigated
Acres
Total
(ac-ft)
per
Irrigated
Acre (ft)
2008
2969.71
9/8/08
2009
2970.78
8/25/09
2010
2971.04
9/5/10
4.3.2. Measurement Comparisons
Overall, the annual water-level measurements and the transducer measurements that have not
been corrected for barometric pressure showed good agreement in the Thomas index well record
(Table 7). In 2008, 2009, and 2010, the discrepancy was 0.06 ft., 0.11 ft., and 0.01 ft.,
respectively.
Year-to-year water-level changes based on the annual well program were -1.4 ft. and +1.4 ft.
between the 07-08 and 08-09 recovery seasons and the 08-09 and 09-10 recovery seasons,
respectively (Table 7). These changes overestimated the water-level decline based on the
maximum recovered water level by 0.9 ft. for the first period, and by 0.2 ft. in the second. It is
noteworthy, however, that both sets of change estimates agreed on the direction of the waterlevel change within each recovery season.
15
Table 7: Annual water-level measurementa comparison with transducer measurements,
Thomas Co.
Date
WL elev (ft)
Indicated Annual
Method
b
WL Change (ft)
1/3/2008
2974.67
NA
Steel tape
c
2974.61
NA
Transducer
1/4/2009
2973.29
-1.38 (-0.45)
Steel tape
c
2973.18
-1.43
Transducer
d
2973.59
NA
Transducer
1/2/2010
2974.64
+1.35 (+1.11)
Steel tape
c
2974.74
+1.56
Transducer
2974.65d
+1.06
Transducer
a
Steel tape measurements are from annual water-level measurement program
(http://hercules.kgs.ku.edu/geohydro/wizard/wizardwelldetail.cfm?usgs_id=383132100543101)
b
Value in () is the decline in the maximum recovered water level measured by the index well
transducer
c
average of values, not corrected for barometric pressure, 0800-1600
d
average of values, corrected for barometric pressure using KGS barometric correction program,
0800-1600
5. Water-level Correction
Significant effort has been expended on correcting water-level measurements recorded by the
pressure transducers in the index wells. Mechanisms that can affect water levels in a well
include changes in barometric pressure, changes in aquifer porosity due to earth tide forces
(stretching and compressing of pores), and major surface loading changes associated with heavy
rainfall and changes in flow in nearby stream channels. In previous reports, earth-tide effects
were shown to have a negligible impact on water levels in the index wells, while the impact of
changes in barometric pressure varied between the index wells. As part of this project, the KGS
has developed an Excel spreadsheet to remove the effect of barometric-pressure fluctuations
from water-level measurements (henceforth, water-level correction). Details and screenshots
from this spreadsheet are available in Appendix C.
5.1.
Water-level Responses to Change in Barometric Pressure
OFR 2010-3 (Buddemeier et al., 2010) provides a detailed explanation of how water levels in
wells respond to fluctuations in barometric pressure, with an emphasis on hydrogeologic
conditions similar to those found in the proximity of the three index wells. We are continuing to
refine our methods for accounting for barometric-pressure impacts on water levels as part of a
complementary research effort. Appendix E contains a paper, which will be published in the
journal Ground Water in 2011, that describes complementary research done at a KGS research
site in the High Plains aquifer in Pawnee County near Larned. The focus of that research is on
assessing the range of hydrogeologic insights that can be gleaned from water-level responses to
16
fluctuations in barometric pressure. Insights developed from that research have been particularly
useful for interpreting the responses of the index wells to barometric-pressure changes.
The deep unconfined aquifer monitored by the Thomas index well displays the largest response
to changes in barometric pressure. As explained in the introduction to the paper in Appendix E,
a change in barometric pressure is instantaneously imposed on the water level in the well.
However, in a deep unconfined aquifer, that change is not immediately imposed on the water
table because of the time needed for the barometric-pressure change to be transmitted through
the vadose zone. This timing difference (barometric lag) between when the well and the aquifer
are affected by the barometric pressure change leads to relatively large water-level changes in the
well. In the Thomas index well, changes in barometric pressure can change the water level by up
to 1.4 ft. in a period as short as three days, even though the actual position of the water table in
the formation has changed very little. The result is a large short-term variation in monitored
water levels, which is easily observed during the recovery season and produces the band in the
Thomas well hydrograph shown in Figure 7. The impact on the year-to-year change estimates
based on the annual water-level measurement program can be large. For example, the annual
water-level measurements in January 2008 and January 2009 were both taken at barometric
pressure extremes (Figure 7, yellow circles). From the index well record, it is clear that
changing the date of either measurement by just ± 3 days could have resulted in estimated annual
water level changes ranging from a 1.4 ft. decline to a 0.3 ft. increase between 2008 and 2009.
This clearly introduces a significant error considering the total water-level variation in this well
is ~6 ft. over the entire record.
To account for the barometric lag between a well and the aquifer, simultaneous water-level and
barometric-pressure measurements must be collected. Barometric pressure measurements were
collected at each index site beginning in January of 2009. For the period prior to that, barometric
pressure information is available from nearby weather stations. When the KGS Excel barometric
pressure correction program (Appendix C) was applied to the data from the Thomas index well,
the water-level uncertainty (the width of the band about the hydrograph) was greatly decreased
(Figure 8).
17
33.0
2977
32.5
32.0
2975
31.5
2974
31.0
30.5
2973
30.0
2972
BP (ft H2O)
Elevation of Water Level (ft ASL)
2976
29.5
2971
Transducer WL
Corrected WL
Corrected WL
Colby BP
Site BP
2970
28.5
1Ju
l- 1
0
30
-S
ep
-1
0
30
-D
ec
-1
0
28.0
1Ju
l- 0
7
30
-S
ep
-0
7
31
-D
ec
-0
7
31
-M
ar
-0
8
30
-J
un
-0
8
29
-S
ep
-0
8
30
-D
ec
-0
8
31
-M
ar
-0
9
30
-J
un
-0
9
30
-S
ep
-0
9
30
-D
ec
-0
9
31
-M
ar
-1
0
2969
29.0
Figure 8: Thomas index well hydrograph (blue line), with corrected water levels
(fluorescent green and turquoise lines). Before January 2009, barometric pressure (BP)
data were only available from a permanent weather station in Colby, KS (olive line).
Correcting water-level measurements for changes in barometric pressure provides several
advantages. Since water-level uncertainty (width of band on hydrograph) is reduced, a clearer
understanding of water-level decline and recovery is provided. This increases confidence in fully
recovered water-level determinations (discussed further in Section 7) and water-availability
estimates. In turn, this enables assessments and decisions to be made sooner, since accumulating
statistics over a number of years is not required.
6. Thomas County Expansion Project
To demonstrate the areal extent over which an index well represents aquifer conditions, KDADWR wells in the vicinity of the Haskell index well were made available to the project.
However, the complex subsurface hydrogeology near the Haskell index well is not representative
of conditions in the High Plains aquifer across most of western Kansas. To address the
representative area issue in more typical High Plains aquifer conditions, additional “wells of
opportunity” were sought in the vicinity of the Thomas index well for transducer installation and
continuous water-level monitoring. An added benefit to monitoring the additional wells (in both
Thomas and Haskell counties) is to provide a proof of concept for the “well of opportunity”
index well monitoring approach detailed by Buddemeier et al. (2010) and in Section 8.
Initially, six wells, including retired and active irrigation wells and a domestic well, were
selected and instrumented with pressure transducers provided by KDA-DWR to monitor the
2009-2010 recovery. Due to sensor malfunction and the desire to enhance data coverage, two
KGS sensors were installed in the fall of 2010. A summary of sensor installation dates and other
significant events is provided in Table 8. Hydrographs from the four monitored wells with
minimal sensor malfunctions are given in Figure 9. Top of casing elevations are currently only
available for TH3 and TH7; the remaining wells will be surveyed shortly. For general
comparison, the well elevations for TH9 and TH10 were estimated using the Google Earth
digital elevation model. Although full recovery information was not available for either TH3 or
18
TH7, water levels in both were clearly higher than in the index well. Water levels in TH10 were
clearly lower than in the index well. This is expected given the water-table map constructed as
part of the Thomas County water budget project indicated an overall west-to-east groundwater
flow field (Figure 10, Appendix D).
Table 8: Installation date and other notes for Thomas Co. expansion wells.
Well
Sensor
Installation Date Notes
TH3
KDA-DWR 8/12/09
Malfunctioned 1/12/10
KGS
9/13/10
TH7
KDA-DWR 9/30/09
Active irrigation well; sensor removed
4/18/10; re-installed 11/23/10
TH8
KDA-DWR 11/5/09
Malfunctioned 12/4/09
TH9
KDA-DWR 11/5/09
Sensor removed 11/11 to 11/14/09 for
well cap installation
TH10 KDA-DWR 8/12/09
TH11 KGS
11/3/10
19
Figure 9: Hydrograph comparison from the Thomas expansion well program utilizing
barometric pressure corrected water levels. Surface elevations for TH9 and TH10 are
estimates using the Google Earth digital elevation model (an elevation survey will be
completed shortly), but are likely accurate to within ±5 ft. The general water-level trend
indicates west-to-east groundwater flow.
20
Figure 10: Groundwater elevation contours near the Thomas index well calculated from
2005 annual measurements (from Buddemeier et al., 2006 – see Appendix D).
Using the water-level data as feet of water above the sensor, some additional observations about
the different wells are possible. TH3, TH7, and TH9 all show trends in water levels that are
similar to that at the index well; when the index well transducer indicated rising water levels, so
did these other three wells. Likewise, when the index well transducer recorded declining water
levels, these other three wells did also. However, the magnitude of these water-level changes
differed, particularly between TH3 and the other wells. With the exception of TH3, hydrographs
indicated 1-3 ft. of recovery, somewhat lower than the recovery recorded in the index well (6 ft.
of recovery). This apparent lower recovery can be attributed to the installation date of sensors
for TH7 and TH9, which occurred at least 30 days after the commencement of recovery. The
first 30 days of recovery are significant, as approximately 50% of the recovery in the Thomas
index well was recorded during this period. Despite installation prior to the end of the pumping
season, the transducer in TH10 only recorded a foot of water-level rise over the entire recovery
period. Once the factors responsible for the small recovery recorded at that domestic well are
clarified, it might prove to be a useful well of opportunity, since it appears subject to very little
pumping perturbation.
The distance to the nearest pumping well also factors into the amount of drawdown and
recovery. The recovery was relatively large at TH3, with water levels rising over 40 ft. prior to
the transducer malfunction. TH3 is a retired irrigation well that is located quite close to its
replacement well. The large water-level changes observed in TH3 are attributed to its immediate
21
proximity to the replacement well. Thus, when pumping ceased in the replacement well, a large
and rapid water-level change in TH3 was observed.
All of the expansion wells displayed pronounced water-level responses to fluctuations in
barometric pressure, consistent with the responses observed at the Thomas index well and the
expected response for this hydrogeologic setting (water table 200+ ft below land surface). The
barometric response functions (BRF) for TH7, TH9, and TH10 indicate an unconfined system
(Figure 11), similar to the Thomas index well (see Figure 3-5, Buddemeier et al., 2010). Using
the BRF, water levels were corrected for barometric response (Figure 12), as in Section 5.1.
Barometric Pressure Response Function
Thomas Expansion Wells
1.20
TH7
TH9
TH10
1.00
Response function (-)
0.80
0.60
0.40
0.20
0.00
-0.20
0
1
2
3
4
5
6
Lag (Days)
Figure 11: Barometric pressure response function for Thomas Co. expansion wells TH7,
TH9, and TH10.
22
Water Level and Barometric Pressure
Corrected WL (ft)
BP (ft)
3015
33
3014
32.5
3013
32
3012
31.5
3011
31
3010
30.5
3009
30
3008
29.5
3007
8/25/09
Barometric pressure (ft water)
Water level elevation (ft)
WL (ft)
29
10/14/09
12/3/09
1/22/10
3/13/10
5/2/10
Time
Figure 12: Well hydrograph, barometric pressure, and water level corrected for
barometric response from Thomas Co. expansion well TH7.
Overall, data from the 2009-2010 recovery provide an initial view of what can be determined
with more complete and extensive monitoring records. If all goes well during the 2010-11
recovery period, more definitive correlations between the index and expansion wells should be
available in mid-2011. Transducers remain to monitor the 2010-2011 recovery, and the
additional data provided will be invaluable for identifying the significance of the intriguing
recovery trends identified in the Thomas index well recovery hydrograph (described below in
Section 7.2).
7. Recovery Water-level Estimation Refinement
In OFR 2010-3 (Buddemeier et al., 2010), three different methods were presented for estimating
the water level at full recovery: a polynomial fit method, a modified Horner recovery method,
and a spreadsheet-based method using the Theis solution (Appendix B in Buddemeier et al.,
2010). The reader is referred to OFR 2010-3 for a more detailed development of the theory and
equations. The Horner recovery method appears to be the most promising of these methods, so a
brief overview is provided here.
In the Horner recovery method, water-level measurements after pumping has ceased (henceforth,
recovery data) are plotted against a time ratio (th), which consists of the duration of pumping (tp)
and the time since pumping ceased (t’), arranged as th = (tp+t’)/t’. For the purposes of this work,
23
the water-level data are in the form of recovery (in ft) since the cessation of pumping, although
the water-level elevation could also be used. The recovery data are plotted against the logarithm
of the time ratio with the ratio decreasing from left to right (t’ increasing). The recovery data are
extrapolated to a th value of one (an infinite time of recovery); that extrapolated value is the
estimated level to which the water would rise at full recovery.
Hydrographs from a few of the wells in the Haskell area indicate water levels actually fully
recover in between pumping periods, and thus can provide proof-of-concept for the Horner
recovery method. This example will focus on HS 20, an irrigation well that has a BRF indicative
of a well screened in an unconfined aquifer. The hydrograph for HS 20 (Figure 13) indicates the
well recovers quickly after cessation of pumping, with pumping for 2007 ending on August 30th.
This was followed by a long period with no pumping, ending on March 20, 2008. Over the
course of the seven-month recovery, water levels were remarkably consistent (Table 9),
indicating water levels reached equilibrium at full recovery. A Horner-recovery plot constructed
for this period indicated a similar recovered water level of 2596.53 ft. AMSL using early
recovery data, and 2596.57 ft. AMSL using later recovery data (Figure 14). This suggests the
Horner recovery method provides viable recovery estimates in the HPA.
HS20 - Horner 07-08
2605
2600
2595
2590
2585
HS20
2580
Early Recovery
Late Recovery
1-Jun-08
12-Apr-08
22-Feb-08
3-Jan-08
14-Nov-07
25-Sep-07
6-Aug-07
17-Jun-07
28-Apr-07
9-Mar-07
2575
Figure 13: HS 20 hydrograph. Blue and pink portions of the recovery were used for the
Horner recovery estimate, and correlate with the early and late recovery period estimates
in Figure 14.
24
Table 9: Average water-level elevation (barometric effects not corrected), HS 20.
Month
Average Water Level (ft AMSL)
Sept. 2007
2596.40
Oct. 2007
2596.65
Nov. 2007
2596.65
Dec. 2007
2596.52
Jan. 2008
2596.54
Feb. 2008
2596.53
Mar 1 – Mar 20, 2008
2596.49
HS20 - Horner 07-08
18
16
14
y = -0.0364Ln(x) + 15.646
12
y = -0.6225Ln(x) + 15.612
h@t'=0=2580.92' AMSL
Horner Recovery = +15.65
Recovered h = 2596.57' AMSL
h@t'=0=2580.92' AMSL
Horner Recovery = +15.61
Recovered h = 2596.53' AMSL
10
8
6
Early Recovery
Late Recovery
4
Log. (Late Recovery)
2
Log. (Early Recovery)
0
1000
100
10
1
Figure 14: HS 20 Horner recovery plot from August 30, 2007, to March 20, 2008, with early
recovery and late recovery estimates. Blue and pink portions of the recovery are the same
as the blue and pink portions in Figure 13.
7.1.
Index Well Horner Recovery Plots
Examples of water-level recovery curves and Horner recovery plots from the Haskell index well
are provided in Figure 15. Panels A and B (Figure 15) use the minimum water level that was
observed immediately before the end of the summer irrigation season, while panels C and D
(Figure 15) use the water level observed immediately prior to the end of the final pumping
period. Dates and water-level elevations corresponding to the end of pumping in Figure 15 are
shown in Table 10. The water-level recovery curves behave as expected (Figure 15A); recovery
progresses quickly for the first ~1000 hrs. At around 700 hours after the minimum water level,
water-level recovery begins to slow down and the recovery curves begin to flatten out. Waterlevel recovery (from minimum water levels) was highest in the 2007-08 recovery and has
decreased in each of the last two years. The Horner recovery plots (Figure 15 B-D) were
25
calculated using an assumed pumping interval (duration of a single irrigation period at a well) of
five days. Using this value and extrapolating recovery curves by hand to a th value of 1, the
estimated final recovery levels for 2007-08, 2008-09, and 2009-10, to the nearest half foot, are
128.5 ft., 125.5 ft., and 121.0 ft. above the minimum water levels (or 2591.0 ft., 2586.5 ft., and
2582.0 ft. above sea level), respectively. This indicates a water-level decline of 4.5 ft. between
2008 and 2009 (between the 2007-08 and 2008-09 recovery seasons), and also between 2009 and
2010 (between the 2008-09 and 2009-10 recovery seasons). These values are consistent with the
annual tape measurements (Table 3), which yield water-level declines of 4.1 ft. and 4.8 ft.
between 2007-08/2008-09 and 2008-09/2009-10, respectively. These values are also consistent
with the declines in the maximum recovered water level measured by the index well transducer
of 5.0 ft. and 3.9 ft. for 2007-08/2008-09 and 2008-09/2009-10, respectively. However, these
values differ significantly from the water-level decline based on the extrapolated quadratic fit
values of 6.4 ft. for 2008-09 (Buddemeier et al., 2010, Table 3.1). While Horner recovery waterlevel values are approximate because of the uncertainty in the assumed pumping interval (five
days) and the manual extrapolations, it is clear that water-level elevations continue to decline in
successive years.
Figure 15: Haskell index well recovery for all three complete recovery seasons (2007-08,
2008-09, 2009-10) and the start of the 2010-11 recovery, plotted as semi-log recovery curves
(A) and as Horner recovery curves (B), (C) and (D). As several pumping events occurred
within the first month after the end of the pumping season, panels (C) and (D) display
recovery from the final pumping event, rather than the lowest recorded water level.
26
Table 10: Comparison of reference water levels used in the Horner plot recovery analysis
for Haskell County. Minimum water level is used in panel B of Figure 15, while the final
pumping period is used in panels C and D.
Minimum Water-Level
2007-08
2008-09
2009-10
2010-11
Date
8/24/2007
8/8/2008
8/16/2009
8/22/2010
ft AMSL
2462.38
2460.84
2460.73
2454.69
Water Level @ end of
final pumping period
Date
ft AMSL
12/7/2007 2567.67
9/27/2008 2553.13
9/8/2009
2514.47
12/24/2010 2562.72
The Haskell index well recovery curves contrast with the two index wells screened in unconfined
portions of the High Plains aquifer (Scott Co., Figure 16, and Thomas Co., Figure 17). The
recovery rate changes for all three index wells around 1000 hrs (~42 days) after the end of the
pumping season, resulting in two distinct recovery periods. However, while the slope of the
recovery curve decreased in the Haskell index well, the slope of the recovery curve increased in
the Thomas and Scott index wells. The difference in responses is primarily a function of the
hydrogeologic setting. In the confined aquifer at the Haskell site, recovery is in its later stages
and water levels are beginning to flatten towards a new equilibrium level. In the unconfined
aquifers at the Thomas and Scott sites, recovery does not appear to “flatten” towards equilibrium.
This difference between the confined and unconfined responses is a function of the difference in
the storage parameters between these two hydrogeologic settings; a confined aquifer has a
storage parameter on the order of 0.0001, while an unconfined aquifer has a storage parameter
(specific yield) on the order of 0.1. The result is a much more rapid recovery of water levels in
confined aquifers. Thus, for the same time since cessation of pumping, the recovery in a
confined aquifer, as a proportion of the total recovery, is much greater than in an unconfined
aquifer.
In the Scott index well (Figure 16), the recovery slopes were consistent, although the slope of
late-time recovery in 2007-08 was lower than the other years, likely due to a shorter recovery
that prevented full expression of the late-time recovery trend. Similar to conditions at the
Haskell site, pumps were turned on several times in the vicinity of the Scott index well during
the recovery in September and October in 2008 and 2009 after the primary irrigation season was
over. This additional pumping affected the recovery at the index well. However, the recovery
curves were consistent from year to year. This consistency indicates that the Horner method
should also be a viable approach for estimating the full recovery level at the Scott index well.
27
Figure 16: Scott index well recovery plotted as semi-log recovery curves with hand-fit
slopes at the late time (A) and as a Horner recovery curve (B) and (C). Two distinct stages
of recovery are clearly evident in all three plots, with an increase shown and clearly
dominant ~1000 hrs from the final pumping period in all three complete recovery seasons
(2007-08, 2008-09, 2009-10) and the start of the 2010-11 recovery.
For the Thomas index well (Figure 17), the slope of the water-level elevation recovery curve
changed consistently in each of the three recovery seasons (2007-08, 2008-2009, 2009-10). The
small slope of the early time recovery (<1000 hrs), which may be the recovery equivalent of the
delayed yield phase of a pumping test in an unconfined aquifer (Batu, 1998) or simply a function
of the specific yield of the aquifer and the distance of the index well from the closest pumping
well, was consistent in each recovery period, as was the much larger slope for the late-time
recovery. This consistency of late-time recovery slopes indicates that the Horner recovery
method should be a viable approach for estimating the level to which the water would rise at full
28
recovery at the Thomas index well. Further assessment of the distinct two-stage recovery curves
at the Thomas and Scott index wells and the factors producing them is presented below.
Figure 17: Thomas index well recovery, plotted as semi-log recovery curves with hand-fit
slopes at the late time (A) and as a Horner recovery curve (B) and (C). Two distinct stages
of recovery are clearly evident in all three plots, with an increase shown and clearly
dominant ~1000 hrs from the final pumping period in all three complete recovery seasons
(2007-08, 2008-09, 2009-10) and the start of the 2010-11 recovery.
29
7.2.
Seeking Equilibrium: Evaluation of Two-stage Recovery and Recovered
Water-level Estimation, Thomas County Index Well
The Thomas index well dataset for the 2009-2010 recovery season (Figure 18) was used for a
detailed comparative analysis, since this is the most complete, uninterrupted dataset available for
any of the sites. The 2009 pumping season got under way on June 24, with pumping ending
August 27, a period of 63 days. Recovery was uninterrupted until June 4, 2010 – a period of 281
days. On August 27, 2009, the barometric pressure corrected (bp-corrected) water-level
elevation was 2971.11 ft.; on June 4, 2010, the bp-corrected water level had recovered to
2976.10 ft. – and had not yet come to equilibrium. The highest observed recovered water level
was 2976.20 ft., on June 21, 2010; just before major irrigation pumping commenced. All
analyses discussed here consider only the time from June 24, 2009, until June 4, 2010.
2977.5
Early Recovery
37.5
36.5
June 24
July 27
35.5
June 4
2974.5
34.5
2973.5
33.5
2972.5
32.5
Aug. 27, 2971.11'
2971.5
31.5
29.5
p10
-S
e
26
-J
un
-M
ar
28
27
-0
9
26
-D
ec
p09
-S
e
26
-J
un
27
-M
ar
-1
0
2969.5
-1
0
30.5
-0
9
2970.5
28
BP (ft H 2O)
2975.5
-0
9
Elevation of Water Level (ft AMSL)
2976.5
Transducer WL
Corrected WL
Site BP
Late Recovery B
Late Recovery B1
Late Recovery B2
Late Recovery B3
Figure 18: Hydrograph of the 2009 Thomas index well pumping season and the subsequent
2009-10 recovery. The graph displays barometric pressure (BP) readings (green line –
reference values on the right y-axis), pressure transducer water-level readings (royal blue
line – reference values on the left y-axis), corrected water-level measurements (bright blue
– references values on the left y-axis), the period used for early-time (short red line) and
late-time trend fits (pink, gold, bright green, and pale blue lines) in Figure 19 and Figure
20. The pink trend fit period encompasses the full length of the gold trend fit, which
encompasses the full length of the bright green fit, which encompasses the full length of the
pale blue fit.
Two scenarios were considered: one where the index well had been influenced only by a single
pumped well for a standard pumping period of five days (Single Well, or SW), and the other
30
where the index well had been influenced by pumping over the entire 63-day irrigation season
(Season Pumping, or SP). This represents an upper limit, as consistent pumping did not start
until July 27. For both scenarios, there is clearly a two-stage recovery. There is no single trend
analysis (linear, log, power, exponential, polynomial) that describes the entire recovery curve
without subdivision. This is the case even when the first two measurements (on the left in Figure
19) are ignored; these two measurements may indicate not all pumps in the area had turned off,
or more likely, represent the early-time confined response typically observed in pumping tests in
unconfined aquifers.
Since the recovery curves in the Horner Plots (Figure 19) appear to have two parts, trend-line
analysis was applied to each part of the recovery. For the earlier, flatter part of the recovery, the
SP and SW analyses indicated that recoveries from maximum drawdown were 1.91 ft. and 1.60
ft., respectively. For the later recovery, the trends indicated recoveries of 6.71 ft. and 5.55 ft.,
respectively.
9
5/1 - 6/4
4/1 - 6/4
3/1 - 6/4
2/12 - 6/4
8
y = -13.891Ln(x) + 7.7963
y = -10.615Ln(x) + 7.1031
7
y = -9.4194Ln(x) + 6.8368
y = -8.851Ln(x) + 6.7055
6
5
y = -88.239Ln(x) + 6.4638
4
3
Recovery Level (ft)
Season Pumping (SP- t'=63d)
Single Well (SW-t'=5d)
SW, early recovery
SP, Early recovery
SW, late recovery
SP, late recovery
SP, late recovery B1
SP, late recovery B2
SP, Late Recovery B3
Log. (SW, late recovery)
Log. (SP, Early recovery)
Log. (SW, early recovery)
Log. (SP, Late Recovery B3)
Log. (SP, late recovery)
Log. (SP, late recovery B2)
Log. (SP, late recovery B1)
2
y = -0.1507Ln(x) + 1.9093
1
y = -0.1733Ln(x) + 1.5988
0
10000
1000
100
10
1
(tp+t')/t'
Figure 19: 2009-10 Thomas index well recovery estimation plot, displaying the differences
between recovery estimates considering the full pumping season (SP, tp=63d) and the
average pumping time of a single well (SW, tp=5d). Recovery estimates for early- and latetime periods are displayed for both the SP and SW calculations.
31
Table 11: Comparison of observed and predicted recovery estimations for the 2009-10
recovery of the Thomas index well.
Recovery Estimates
Initial WL
(ft.)
2971.11
2971.11
Annual Program
Max. Observed
(transducer)
Whole Pumping Season (SP)
Early time
2971.11
Late time “B” (2/12-6/4) 2971.11
Late time “B1” (3/1-6/4) 2971.11
Late time “B2” (4/1-6/4) 2971.11
Late time “B3” (5/1-6/4) 2971.11
Single Well (SW)
Early time
2971.11
Late time
2971.11
Recovered WL
(ft).
2973.29
2976.20
Indicated Recovery
(ft.)
2.18
5.09
2973.02
2977.82
2977.95
2978.21
2978.91
1.91
6.71
6.84 (↑ 0.13)
7.10 (↑ 0.26)
7.80 (↑ 0.70)
2972.71
2977.57
1.60
6.46
It is clear that the early-time trends do not represent full recovery. All late-time scenarios
indicated recovered water levels higher than the highest observed water level (2976.20 ft.) and
much higher than the early trend (2972.71 ft./2973.02 ft.) and annual program (2973.29 ft.)
estimates. However, these late-time estimates are themselves lower limits on the probable
recovered value. During trend analysis of the SP scenario, several late time periods were
considered – one from February 12 to June 4, one from March 1 to June 4, one from April 1 to
June 4, and a final one from May 1 to June 4. Each later time period provided increased
estimates of the predicted recovered water level (Table 11 and red line in Figure 20). The
increasing rate of recovery estimates considering later time periods indicates that late recoveries
are still clearly trending upward even at the endpoint suggested by the original analysis. Based
on the curves in Figure 18, it seems likely that full recovery could require a year or more.
Characterizing the recovery periods, amounts, and time constants is a first step toward
understanding their causes. It seems likely that the two-stage recovery plots result from the
unconfined nature of the aquifer in the vicinity of the Thomas index well. However, different
portions of the aquifer will influence different portions of the recovery process so heterogeneity
in aquifer conditions may also be playing a role. The late-time recovery pattern is most likely
just the typical late-time recovery behavior observed in unconfined aquifers where specific yield
is the appropriate storage parameter. However, the response could be interpreted to indicate that
a recharge boundary has been reached at some distance from the index well.
In the case of the Thomas site, the second possibility cannot be ruled out. Water budget studies
in 2005-2006 showed that the index well site is only 4-5 miles NE of an undeveloped aquifer
fringe zone with higher water-table elevations and gradients, and with a flow direction generally
trending NE (Figure 10, Appendix D). Several irrigation wells to the northeast of the index well
have also been retired, with land use converted to dry-land farming, over the course of the
project. Quantitative studies will be required to separate this mechanism from the typical
unconfined aquifer response or lateral recharge from the stream channel to the north (which,
32
although usually dry, serves as a water collector in wet years). There are data available to
support at least preliminary modeling to test this lateral recharge hypothesis, and the Thomas
expansion wells provide locations at which differences in the timing and rate of recovery can be
used to develop inferences about the water source(s) driving the late recovery.
9
Season Pumping (SP- t'=63d)
SP, late recovery B1
SP, late recovery B2
SP, Late Recovery B3
Log. (SP, Late Recovery B3)
5/1 - 6/4 y = -13.891Ln(x) + 7.7963
8
4/1 - 6/4 y = -10.615Ln(x) + 7.1031
7
3/1 - 6/4 y = -9.4194Ln(x) + 6.8368
2/12 - 6/4
6
y = -8.851Ln(x) + 6.7055
Log. (SP, late recovery)
5
Log. (SP, late recovery B2)
Log. (SP, late recovery B1)
4
3
Recovery Level (ft)
SP, late recovery
2
1
0
10
(tp+t')/t'
1
Figure 20: 2009-10 Thomas index well recovery estimation plot, displaying the differences
between recovery estimates considering data from differing late time periods. The higher
(tp+t’)/t’=1 intercept for later time period data indicates the slope of the late time data is
still increasing.
Recovery estimates for the Thomas index well were prepared from each recovery period (Table
12). Pumping occurred during the 2007-08 and 2008-09 recovery, not as part of the main
pumping season, but lasting for a clearly defined time-period. The recovery period after these
pumping events did not last for sufficient time to observe the two-stage recovery. In 2007-08,
recovery analysis of these late-period pumping events provided roughly similar water-level
recovery estimates as from the late periods of the main recovery (considering only the single
well pumping event for the main recovery season).
33
Table 12: Summary of recovered water-level estimates and observed values, Thomas index
well. Graphs of each recovery period are available in Appendix B.
Recovery
Year
Recovery
Period
2007-08
Main
(Start 9/8/07)
Period 2
(Start 5/21/08)
ht’=0 (ft)
Recovery
(ft)
Recovered
Elevation
(ft AMSL)
Fit Period
Largest t'
value (hrs)
2970.53
5.01
2975.54
Observed
5933
60?
6.00
2976.53
5
5.71
2976.24
1.82
2975.47
Observed
2.58
2.65
2976.23
2976.30
225-476 hrs
225-476 hrs
4.94
2975.09
Observed
118
6.26
2976.41
5
5.72
2975.87
2.00
2974.95
Observed
2.31
2975.26
659-1395 hrs
1.31
2975.09
Observed
1.43
2975.21
242-532 hrs
5.09
2976.20
Observed
63
6.71
2977.82
63
6.84
2977.95
63
7.10
2978.21
63
7.80
2978.91
5
6.46
2977.57
2.22
2973.30
Observed*
(Nov 2010)
2.73
2.26
2973.81
2973.34
218-394 hrs
159-683 hrs
tp
(d)
2.83
2973.65
2.83
5
2008-09
Main
(Start 9/14/08)
Period 1
(Start 3/23/09)
2970.15
7.54
2972.95
7.54
Period 3
(Start 6/2/09)
2.67
2973.78
2.67
2009-10
2010-11
Main
(Start 8/27/09)
2971.11
Main
(Start 9/6/10)
2971.08
77
5
1883-5933
hrs
1883-5933
hrs
476
4382
2817-4382
hrs
2817-4382
hrs
1395
532
7163
4077-7163
hrs
4466-7163
hrs
5209-7163
hrs
5943-7163
hrs
4077-7163
hrs
1362*
(Nov 2010)
*2010-11 recovery is ongoing. As described in the text, early-time recovery estimates in Thomas
County underestimate recovered water levels.
34
7.3.
Summary: Recovery Water-level Estimation
The difference in recovery plots between index wells is primarily a function of the hydrogeologic
setting: unconfined (water-table or phreatic) conditions versus confined conditions. The Scott
and Thomas index wells are screened in unconfined aquifers, whereas the Haskell index well is
screened in a confined aquifer. The hydrogeologic setting determination was based on a
hydrostratgraphic analysis of the drillers’ logs, consideration of water-level responses to
pumping and the cessation of pumping, and an assessment of water-level responses to
fluctuations in barometric pressure.
The numerous monitored wells in the vicinity of the Haskell site provide an opportunity to gain
insight into the nature of the late-time change in recovery rate. Appendix A illustrates the range
of responses observed at those wells during the 2007-08 recovery period. Although an
assessment of the data in this and subsequent recovery periods is ongoing, some initial
observations can be made. Using data from wells where there was a definable minimum water
level at the end of the irrigation season and a complete recovery record, a change in recovery
slope was observed ~1000 hrs into the recovery. Unlike at the Scott and Thomas index wells,
however, the semi-log rate of recovery decreased in virtually all of the wells, as the recovery
began trending towards a definable recovered water level. However, there are indications that
exceptions to this behavior exist as the raw data from Haskell are corrected for barometric
pressure response. Overall, the data indicate that the increasing rate of recovery observed in the
Thomas and Scott index wells was not observed in the Haskell site wells at the same time since
cessation of pumping. The exact significance of this observation has yet to be determined, but it
most likely is a function of the large difference in storage parameters between confined and
unconfined aquifers. However, the phenomenon might also be partly due to some additional
source of water that must travel a greater distance to the well. The additional monitoring
locations in Thomas County should provide more insight into these differences in recovery
during the 2010-2011 recovery season. Regardless, the observations to date suggest that the twostage recovery is associated with unconfined aquifers.
The two-stage recovery process observed in the Scott and Thomas index wells has a number of
implications for interpretation and management. It helps to explain some of the variations noted
in recovery estimations based on quadratic curve fits (Buddemeier et al., 2010), and it suggests
that greater precision and confidence can be achieved when that late-time recovery slope is
reached. At this time, although further refinement of the technique is necessary, the recovered
water-level estimates gained from this procedure for the Thomas and Scott index wells should be
considered a minimum value. Further analyses are required to assess how much these values
underpredict the water level at full recovery. Despite the continued uncertainty in estimates of
recovered water level at these sites, the monitored index well approach continues to provide
substantial improvement over once-a-year water-level monitoring for enhanced management.
One additional question that has yet to be resolved is the influence of spatial variations in
pumping on estimates of water level at full recovery. For example, will a well in a lightly
pumped area surrounded by regions of greater annual drawdown likely yield a greater full
recovery water level than nearby wells that are in more heavily pumped areas? Theory predicts
that variations in pumping will not influence the fully recovered water level if the drawdown
35
cones interact during the pumping period. However, the issue becomes more complicated if the
drawdown cones do not interact. Answering such questions is vital to understanding the
effectiveness of management strategies and resource sustainability on the sub-unit scale, and will
become possible as more data from the expanded Thomas County study area are obtained
(Section 6). The zone of influence (area within which drawdown is >0.1 ft.) of a 1000 gpm well
in a medium- or high-transmissivity aquifer has been estimated as about 1.5 miles (Buddemeier
et al., 2002, p. 21 Fig. 6), and the Thomas index well has 12-15 pumping wells within that
approximate radius. Thus, it is doubtful if spatial variations in pumping will greatly influence
estimates of fully recovered water level in the vicinity of the Thomas County study area.
8. Well Hydrographs and Use of Wells of Opportunity
As described in past reports (Young et al., 2008, Buddemeier et al., 2010), geology is an
important consideration when interpreting and comparing water-level measurements. Variations
in the thickness and distribution of geologic strata affect both the water level and the response to
barometric pressure changes. Spurred on by the findings described in Section 7, efforts this year
have focused on identifying and understanding how geology and well construction affect
hydrographs in specific wells. Here, the additional water-level data collected in active and
retired irrigation wells, observation wells, and domestic wells in Haskell and Thomas counties
are examined to identify important characteristics of well hydrographs. The goal of this
examination is to improve understanding of the areal extent over which findings from an index
well are applicable, and to identify ideal characteristics for wells of opportunity for verifying
index well data.
A number of different well responses were observed in both Haskell and Thomas counties. In
some wells, water levels only recover a few feet after reaching their minimum water level (after
a small maximum decline in water level during the irrigation season). In other Haskell County
wells, however, water levels recover upwards of 130 ft. from their minimum levels. A number
of Haskell DWR wells exhibit long, slow water-level decline, with minimal or no recovery.
Finally, hydrographs from a separate group of pumping wells indicate a low efficiency or low
transmissivity in the immediate vicinity of the well. These various categories are discussed
further below.
Small water-level declines and subsequent recoveries (<10 ft.) are typically observed in wells
located at some distance from pumping wells in unconfined aquifers. These will be the most
common type of hydrograph from index wells located in the High Plains aquifer, and will also
include other monitoring and retired irrigation wells. Examples include the Thomas and Scott
index wells (Figure 5 and Figure 7), wells TH 7, 9, and 10 (Figure 9 and Figure 11), and wells
HS 5, 6, 8, 12, and 28 (see Appendix A.4 in Buddemeier et al., 2010 for Haskell County well
hydrographs and BRFs).
Significant water-level declines may be observed in inactive or active wells screened in confined
aquifers, or in unconfined aquifers where the well is located near the pumping well (or is the
pumping well, particularly in areas of low transmissivity or in cases of low well efficiency; the
latter cases will be discussed further below). In the Haskell area, there are several examples of
36
the former case, including the index well (Figure 3) and HS 1, 2, 18 (see Appendix A.4 in
Buddemeier et al., 2010, for Haskell County well hydrographs and BRFs). TH3 in Thomas
County (Figure 9) is an example of the latter case. In the absence of a well log, the difference in
hydrogeologic setting between these cases can often be identified with the barometric response
function (see Section 3.3 in Buddemeier et al., 2010 for a further discussion of barometric
responses typical for confined and unconfined settings).
The group of wells with long, slow declines and minimal recovery consists entirely of retired
irrigation wells found in Haskell County (HS 3, 10, 13, 14, 15, 17, 30; hydrographs are available
in Appendix A.4 in Buddemeier et al., 2010). The BRFs for these wells indicate they are all
screened in the unconfined portion of the aquifer – which is expected for retired irrigation wells
in Haskell County. Water levels in these wells begin a long, slow decline over the course of the
irrigation season. Nearly half of the wells decline beyond the end of the irrigation season into
October or even early December, with little to no subsequent recovery. In all wells, declines
were equal to ~5 ft. in 2007 and 3-4 ft. in 2008. The cessation of water-level declines roughly
coincides with the late-time (1000-2000 hrs after minimum water level) reduction in the rate of
water-level recovery in the Haskell index well. The similarity of water-level declines recorded
in wells located within 2.5 linear miles of each other is promising. It is not clear at this time why
water levels in these retired irrigation wells respond in such a delayed fashion. Although the
screen location of these wells is in the unconfined aquifer whereas the index well is screened in
the confined aquifer, a similar water-level decline is indicated by both the retired irrigation wells
and the index well. This agreement is most probably due to water transmission through the clay
layer separating the two aquifers at the Haskell site. Despite the unknowns regarding the waterlevel response, these hydrograph records still appear useful for determining recovered water
levels.
Finally, hydrographs from several of the active irrigation wells indicated they could be operating
at a low efficiency (e.g., HS 9, 11, 20, 21, 29, and 30; hydrographs are available in Appendix A.4
in Buddemeier et al., 2010). Hydrographs from these wells are characterized by water-level
drops of >20 ft. immediately after pumping is initiated at the well, followed by an immediate rise
in water level of nearly the same magnitude, although typically slightly less, once pumping
ceases. Over the recovery season, water levels further rise another 1-2 ft.. Previous KGS work
(Butler, 1988) has established that well inefficiency will not influence late-time responses, so
these wells should still provide useful measures of water-level recovery for a sub-unit area.
9. Temporal and Regional Trends: Water Levels and Water Use
Quantifying the relationship between water use and recovered water levels at the three index well
sites is an important part of this study. At this time, data are insufficient to draw conclusions
with any statistical confidence, but it is worth noting that water use and recovered water levels
are following expected trends and early indications suggest they are correlated (Table 13). At
the Scott and Haskell sites, water levels decreased during both the 2008-09 and 2009-10
recoveries compared to the previous year. However, in both cases, the decline was less in 200910 than in the previous year, and was correlated with a reduction in water use in 2009 compared
with 2008. In this limited dataset, water use appears to have had a direct impact on the recovered
37
water levels – both observed and predicted. The magnitude of water-level change related to a
change in water use will vary from site to site, based on local conditions including geology. At
the Thomas site, water use within a 2-mile radius of the index well was ~900 ac-ft lower in 2009
than in 2007 or 2008. This resulted in an increase in both the observed and predicted recovered
water levels. Although the increase could be a product of a longer recovery period (observed) or
an inappropriate extrapolation to full recovery (predicted), it also could be providing important
information about the behavior of the High Plains aquifer in the vicinity of the Thomas site. For
example, if all the extracted water was being “mined” from the aquifer in the vicinity of the
Thomas site, water-level declines, albeit smaller, should still have been observed. As water-level
recovery estimates improve, and additional data become available, the relationship between
water use and recovered water levels should provide a valuable resource for management on the
subunit aquifer scale.
Table 13: Water use and recovered water levels at the three index well sites.
Water
Usea
ac-ft
∆ WU
Hmaxb
observed
ft AMSL
∆ WL
Hmaxb
predicted
ft AMSL
ac-ft
ft
2007 (07-08
2868.87
-747.01
2975.54
NA
2976.53
Recovery)
2008 (08-09
2825.21
-43.66
2975.09
-0.45
2976.41
Recovery)
2009 (09-10
1917.17
-908.04
2976.20
+1.11
2978.91
Recovery
Scott
2007 (07-08
3095.78
-564.12
2835.9
NA
2836.26
County
Recovery)
2008 (08-09
4014.33
+918.55
2834.7
-1.2
2835.04
Recovery)
2009 (09-10
2955.48
-1058.85
2834.2
-0.5
2834.61
Recovery
Haskell
2007 (07-08
8764.01
-540.01
2586.1
NA
2587.03
County
Recovery)
2008 (08-09
9931.71
+1167.7
2581.1
-5.0
2581.64
Recovery)
2009 (09-10
8720.45
-1211.26
2577.2
-3.9
2576.71
Recovery
a
– within a 2-mile radius of the well.
b
– Hmax = maximum recovered water level; for predicted, using pumping season for tp.
Thomas
County
∆ WL
ft
NA
-0.12
+2.50
NA
-1.22
-0.43
NA
-5.39
-4.93
Determining the area represented by the recovered water-level estimates is also important to the
success of the index well approach to aquifer subunit management. The area that water-level
change in an index well represents is considered dependent on local conditions, including
geology and spatial water use patterns. As the water table or piezometric surface elevation can
vary considerably across a small area (as at the Thomas site, see Figure 10 and Figure 9), the
recovered elevation is not as important for determining the representative area for any given
index well as is the year-to-year change in recovered elevation.
The Haskell site, with multi-year hydrographs available from 20 wells near the index well (see
Appendix A.4, Buddemeier et al., 2010, for details of each well) currently provides the only
opportunity to investigate the representative area of the index well. In the Haskell area, the
analysis is complicated by the subsurface geology, with wells screened in an unconfined or
38
confined aquifer, or across both. The approach in this case is to compare changes in water level
in wells with similar barometric response functions, indicating they are screened in similar
aquifer units. The results of the recovery analysis for the Haskell wells are shown in Table 14,
with a statistical summary provided in Table 15. None of the hydrographs in the Haskell area
have been corrected for barometric lag, and a large amount of noise was evident in many of the
hydrographs. Both of these issues disproportionately affect recovery estimations for wells
screened in the unconfined aquifer. As such, there is an uncorrected error present in the
calculations. Furthermore, many of the wells were not designed as monitoring wells, and are
screened across large portions of the aquifer. In these wells, water levels represent an integrated
value across the aquifer, and may not represent the true water table or the piezometric surface for
the confined aquifer. The average year-to-year changes in water level for wells screened in the
confined aquifer (Table 14 and Table 15) are within one standard deviation of the 4.5 ft. yearly
change in estimated recovered water level at the index well. This agreement indicates that the
index well provides a good indicator of water-level change in the confined portion of the aquifer
in the Haskell area. If an index well were installed in the unconfined aquifer in the Haskell area,
the analysis indicates that it would also provide a good indicator of water-level change over the
area in question. Over the next year, the recovered water-level data will be refined as the
hydrographs in the Haskell area are corrected for barometric lag. Complimentary data are also
available from Rawlins and Stevens counties (Section 10.1), and with the addition of the Thomas
county expansion wells (Section 6), additional insight into area of response similarity will be
available in the next year.
Table 14: Water use and provisional recovered water levels in wells near the Haskell
County index well.
HS
Type
Aquifer
Year
2 mi.
Water
Use
ac-ft
1
Irrigation
Conf.
07-08
08-09
09-10
07-08
08-09
09-10
07-08
08-09
09-10
07-08
08-09
09-10
07-08
08-09
09-10
07-08
08-09
09-10
07-08
08-09
8756.78
8406.61
8519.90
8723.78
8433.89
8901.90
9081.78
8762.89
9243.90
6998.14
7264.19
7139.45
8764.01
9931.71
8720.45
8443.78
8147.61
8768.90
7145.86
7284.20
2
Irrigation
Conf.
4
Monitoring
Conf.
18
Irrigation
Conf.
Index
3
5
Monitoring
Irrigation (ret)
Monitoring
Conf.
Unconf.
Unconf.
39
Hmax
(Horner)
∆ WL
year/year
∆ WL
(07-08 to
09-10)
ac-ft
ft. AMSL
ft.
ft.
-617.25
-350.17
+113.29
-529.25
-289.89
+468.01
-685.26
-318.89
+481.01
-1364.28
+266.05
-124.74
-540.01
+1167.70
-1211.26
-479.25
-296.17
+621.29
-1028.36
+138.34
2591.32
2584.61
-6.71
∆ WU
2589.67
2585.98
2583.24
n.d.
2582.47
2580.66
2593.09
2586.73
2583.13
2587.03
2581.64
2576.71
2590.85
2587.14
2583.13
2592.87
2589.01
-3.69
-2.74
-6.43
-1.81
-6.36
-3.60
-9.96
-5.39
-4.93
-10.32
-3.71
-4.00
-7.71
-3.86
HS
6
7
8
9
10
11
12
13
14
Type
Monitoring
Irrigation
Monitoring
Irrigation
Irrigation (ret)
Irrigation
Irrigation (ret)
Irrigation (ret)
Irrigation (ret)
Aquifer
Unconf.
?Unconf.
Unconf.
Unconf.
Unconf.
Unconf.
Unconf.
?Unconf.
Unconf.
15
Irrigation (ret)
Unconf.
20
Irrigation
Unconf.
21
Irrigation
Unconf.
29
Irrigation
Unconf.
31
Irrigation
Unconf.
Year
2 mi.
Water
Use
09-10
07-08
08-09
09-10
07-08
08-09
09-10
07-08
08-09
09-10
07-08
08-09
09-10
07-08
08-09
09-10
07-08
08-09
09-10
07-08
08-09
09-10
07-08
08-09
09-10
07-08
08-09
09-10
07-08
08-09
09-10
07-08
08-09
09-10
07-08
08-09
09-10
07-08
08-09
09-10
07-08
08-09
09-10
Hmax
(Horner)
ac-ft
ac-ft
ft. AMSL
ft.
ft.
7515.29
8397.86
8179.64
8390.36
7378.93
7313.22
7444.74
7126.21
7483.61
7444.15
10468.46
10793.43
11118.67
8090.95
7922.55
8651.65
8104.91
7946.61
8528.99
7380.58
7319.37
7437.88
6374.29
6431.40
6511.7
8287.79
7822.50
8070.03
7335.78
7374.89
7499.90
6816.20
7163.61
7174.15
6423.20
6772.61
6591.15
9631.73
9624.37
10095.69
8130.44
7821.54
8072.52
+231.09
-191.05
-218.22
+210.72
-1222.16
-65.71
+131.52
-1095.67
-39.46
+357.40
-1543.93
+324.97
+325.24
-1048.69
-168.4
+729.1
-1259.96
-158.30
+582.38
-995.51
-61.21
+118.51
-873.66
+57.11
+80.3
-207.96
-465.29
+247.53
-692.26
+39.11
+125.01
-1168.67
+347.41
+10.54
-1134.67
349.41
-181.46
-855.84
-7.36
+471.32
-877.98
-308.90
+250.98
2584.19
2589.39
2584.86
2580.54
2593.96
2587.64
2584.07
2597.22
2592.93
2589.38
2591.13
2587.54
2583.59
2595.20
2591.01
2587.10
2596.46
2592.30
2588.53
2593.02
2589.15
2585.90
2593.25
2590.80
2585.00
2589.75
2586.10
2585.00
2592.03
NA
2583.31
2597.08
2592.30
2588.71
2597.56
2592.91
2589.14
NA
2583.28
2576.79
2597.35
2593.41
2589.86
-4.83
-8.68
-4.53
-4.32
-8.85
-6.32
-3.57
-9.89
-4.29
-3.56
-7.84
-3.59
-3.95
-7.54
-4.19
-3.91
-8.10
-4.16
-3.77
-7.93
-3.87
-3.25
-7.12
-2.45
-5.80
-8.25
-3.65
-1.10
-4.75
40
∆ WL
year/year
∆ WL
(07-08 to
09-10)
∆ WU
-8.72
-4.78
-3.60
-8.38
-4.65
-3.77
-8.42
-6.49
-3.94
-3.55
-7.49
Table 15: Statistical summary of water use and recovered water level from Table 14 for
wells in the vicinity of the Haskell index well.
Aquifer
Year
∆ WU
s.d.
Confined Aquifer
07-08
-716.01
364.26
08-09
09-10
99.85
647.65
07-08
-903.43
401.55
08-09
09-10
-24.67
289.57
Unconfined Aquifer
∆ WL (Horner)
s.d.
-5.54
-3.27
1.35
1.32
-4.14
-3.96
0.85
1.20
10. Spin-offs and Related Research
As the Index Well Project progressed through the fourth year, several complementary efforts
developed to further the work of the project.
10.1.
Rawlins and Stevens Counties
KDA-DWR has supplied the KGS with several years of water-level data recorded by pressure
transducers at sites in Rawlins and Stevens counties. These data are being processed and
analyzed with the techniques developed in the course of this project. This will allow the KGS to
further test the usefulness of wells of opportunity, explore aquifer similarities and differences at
additional locations, and enhance confidence in the techniques developed to date.
10.1.1.
Rawlins County
The KDA-DWR site in Rawlins County consists of three wells in section 25 of 3S 36W. The
well designated 23289 is an active irrigation well (depth 253 ft.) in the NW quarter of the
section, obs23289 is an observation well (depth = 250 ft.) in the SW quarter, and obs28290 is an
observation well (depth = 252 ft.) in the NE quarter (http://abyss.kgs.ku.edu/pls/abyss/wwc5).
The observation wells are each several hundred yards away from 23289, and obs28290 is close
to another irrigation well (28290). The two observation wells are nearly a mile apart.
Hydrographs displaying the available records for all three wells are shown in Figure 21. Other
than the expected differences in drawdown between the pumping and observation wells, the
hydrographs are very similar in overall form.
41
Figure 21: Hydrographs of wells 23289, obs23289, and obs28290 in sec. 25, T. 3 S., R. 36
W., Rawlins Co.
Water levels in all three wells were still rising when pumping resumed in May of 2008 and 2009,
and it appears that the fully recovered water levels would be a significant fraction of a foot lower
in 2009 than in 2008. The records have not yet been analyzed for the two-stage recovery.
The 2007-2008 recovery period in all three wells was analyzed using the barometric response
correction spreadsheet tool. Figure 22 - Figure 23 compare the water-level correction plots and
the BRF plots for the same period in each of the wells.
42
Figure 22: Measured water levels, barometric pressures, and corrected water levels for
wells 23289, obs23289, and obs28290 in sec. 10, T. 3 S., R 36 W., Rawlins Co.
43
Figure 23: Barometric response functions for wells 23289, obs23289, and obs28290, in sec.
10, T. 3 S., R 36 W., Rawlins Co.
44
The three water-level recovery curves, both corrected and uncorrected, are very similar, as are
the three barometric response functions. Additionally, the 07-08 and 08-09 recovery periods in
the well 23289 hydrograph were compared; as expected, the BRF plots from those two periods
were essentially identical. In all cases, the BRF plots indicate that the Rawlins Co. wells are
screened in an unconfined aquifer overlain by a thick vadose zone.
These results indicate that the aquifer characteristics vary only to a very minor degree over the
section-level range of this study, and show that for this location and scale, the index well concept
is very well supported. Even more significantly, the results are very similar to those observed at
the Thomas index well, approximately 30 miles to the south. This provides reason to be hopeful
that index wells in the northern (GMD4) section of the HPA would be representative of
conditions over relatively large geographic areas.
10.1.2.
Stevens County
Data from the Stevens County site consist of records from six different wells, located in secs.
4,5,6, 8, and 9, T. 35 S. R. 36 W. Portions of the available hydrographs for these wells are
shown in Figure 24. The difference between maximum and minimum water levels in all wells
was more than 150 ft. and the difference was greater than 200 ft. in three of the wells. These
differences are even larger than those observed in the confined aquifer at the Haskell County site
(Figure 3), which strongly suggests that the Stevens County wells are also screened in a confined
or semi-confined aquifer.
A preliminary check of the barometric response functions has been made, and all appear to be
similar to those from the Haskell Coounty index well (Buddemeier et al., 2010, Figure 3-5). This
also supports the confined aquifer interpretation, but additional analysis of the barometric
response functions is needed.
45
Figure 24: Hydrographs for wells 42421, 42423, 42453, 40578, obs40578, and 44593 in
Stevens County. Note that the two graphs do not cover the same time period; the upper
plots begin shortly before the lower ones end.
10.2.
Haskell County NSF Project
In the summer of 2010, the KGS was awarded a $381,000 grant from the National Science
Foundation (NSF) to study the subsurface stratigraphic framework, sedimentary facies, and
chronostratigraphy of the Ogallala Formation and overlying units. Haskell County will be the
46
focus of this investigation. At least one of the boreholes drilled during this study will be located
adjacent to the Haskell County index well. If possible, as part of this complementary project, a
second monitoring well will be completed at the Haskell site just above the clay confining unit.
This would enable a more controlled comparison with the index well data and the development
of a better understanding of recovery responses in the unconfined aquifer at the Haskell site.
10.3.
Department of Energy Grant – NMR Investigations of Index Wells
In the fall of 2010, the KGS was awarded the first phase ($21K) of a grant subcontract from the
Department of Energy to work together with Vista Clara, a company located near Seattle,
Washington, and Stanford University on assessing the potential of nuclear magnetic resonance
(NMR) technology for estimation of water-filled porosity and permeability using small-diameter
(2-5” ID) wells. Although still in the development and testing stages, the new NMR tool was
applied to the Thomas and Haskell index wells in the late fall of 2010 to develop a better
understanding of the hydrostratigraphy at each site. The Scott site was not selected due to
concerns about electrical interferences from the nearby radio tower. Data from these tests are
still preliminary and are currently being analyzed.
11. Summary and Conclusions
We now have collected hourly water-level data from each of the three index wells for four years;
data are available publicly online via satellite telemetry. Additional water-level data have been
collected from nearby wells in Haskell and Thomas counties, and from two additional groups of
wells in Rawlins and Stevens counties. This large body of water-level data has increased our
confidence in water-level results from the index wells and has provided the opportunity to
demonstrate the utility of an index well for improving estimates of water-level change in an
aquifer sub-unit area.
Through this dataset, it has become clear that, for a variety of reasons, measurements collected
during the annual program do not provide full recovery estimates. In fact, water levels in most
of the wells do not recover prior to the start of the following pumping season. Thus, a major
focus of the project has been on the development of methods to estimate the elevation to which
water levels would recover in the absence of further pumping. This year, rigorous approaches
for extrapolation of water levels to full recovery were developed further. These techniques have
already increased understanding of recovery characteristics, providing information critical to the
accurate determination of changes in water in storage, and therefore to the evaluation of the
effectiveness of any enhanced management procedures. The two-stage recovery process
identified in the Scott and Thomas index wells potentially has a number of implications for
interpretation and management.
The data from all of the wells show that water-level measurements are affected by changes in
barometric pressure, thus simply monitoring water levels was not sufficient to accurately
determine recovered water levels. At sites similar to the Thomas County index well, changes in
barometric pressure alone can produce changes in water level exceeding a foot. Thus, a major
47
focus of the project has been on the development of methods to remove the impact of barometric
pressure changes from the water-level data (i.e. “correct” the water-level data). A spreadsheet
was developed (Appendix C) for that purpose. A manuscript was completed in 2010 and will be
published in 2011 on complementary work at the KGS Larned Research Site (Appendix E). This
complementary work has been extremely valuable for the interpretation of index well responses
to fluctuations in barometric pressure.
A key question concerns the areal reach of an index well, i.e. how broadly, in a geographic sense,
can the findings from an index well be applied. In order to address that question, water-level
data from additional wells in Haskell, Rawlins, and Stevens counties were made available by
KDA-DWR, and additional wells of opportunity were instrumented in Thomas County. The data
from Haskell County demonstrate the index well is representative of local conditions and have
enhanced our overall understanding of the subsurface in that area. Similarities of responses in
Thomas and Rawlins counties and Haskell and Stevens counties point to the encouraging
potential for index well application at other sites with either confined or unconfined aquifers.
These observations indicate that future index well installations can be calibrated for their areal
reach by acquiring relatively short-term additional pressure transducer data from nearby wells of
opportunity (at least five additional wells for one to two recovery seasons).
In the fifth year of this project, we will primarily focus on the following five activities:
1) Complete processing of all the water-level data from the three index wells, the Haskell and
Thomas expansion wells, and the additional wells in Rawlins and Stevens counties;
2) Finalize the approaches for extrapolation of water levels to full recovery;
3) Further our understanding of the relationship between changes in the estimated water level at
full recovery and water use at the index wells and auxiliary sites;
4) Provide more definitive calibration of the areal reach of each index well; and
5) Develop procedures for incorporating uncertainty produced by barometric pressure into
annual water-level survey measurements.
In the report of the fifth year of the project, we will summarize the major findings of the project
to date and suggest avenues for expanding the index well approach to elsewhere in the High
Plains aquifer and for incorporating the approach directly into practical management and
assessment activities.
48
12. References
Batu, V. 1998. Aquifer hydraulics. John Wiley & Sons, New York, 752 pp.
Buddemeier, R.W., R. Stotler, J.J. Butler, Jr., W. Jin, K. Beeler, E. Reboulet, P.A. Macfarlane, S.
Kreitzer, D.O. Whittemore, G. Bohling, and B.B. Wilson. 2010. High Plains Aquifer
Calibration Monitoring Well Program: Third Year Progress Report. Kansas Geological Survey
Open-File Report 2010-3. Available online at:
http://www.kgs.ku.edu/Hydro/Publications/2010/OFR10_3/index.html
Buddemeier, R.W., B.B. Wilson, J. Mosteller, and G. R. Hecox. 2002. Scale, uncertainty,and the
relationship between basic data, information, and management perspectives. Kansas
Geological Survey Open-File Report 2002-25F. Available online at:
http://www.kgs.ku.edu/HighPlains/OHP/2002_25F.pdf
Buddemeier, R.W., D. Young, and B.B. Wilson. 2006. Water Budgets, 4-township Thomas Co.
Region. Draft letter report prepared for GMD4. (Please see Appendix D).
Butler, J.J., Jr. 1988. Pumping tests in nonuniform aquifers - the radially symmetric case. Journal
of Hydrology, v. 101, no. 1/4, 15-30.
Hecox, G.R., P.A. Macfarlane, and B.B. Wilson. 2002. Calculation of yield for High Plains
aquifer wells: relationship between saturated thickness and well yield. Kansas Geological
Survey Open-File Report 2002-25C.
Massuel, S., J. Perrin, M. Wajid, C. Mascre, and B. Dewandel. 2009. A simple, low-cost method
to monitor duration of ground water pumping. Ground Water, v. 47, no. 1, 141-145.
Young, D.P., R.W. Buddemeier, D.O. Whittemore, and E. Reboulet. 2007. High Plains Aquifer
Calibration Monitoring Well Program: Year 1 progress report on well installation and aquifer
response. Kansas Geological Survey Open-File Report 2007-30.
Young, D.P., R.W. Buddemeier, J.J. Butler, Jr., W. Jin, D.O. Whittemore, E. Reboulet, and B. B.
Wilson. 2008. High Plains Aquifer Calibration Monitoring Well Program: Year 2 progress
report. Kansas Geological Survey Open-File Report 2008-29.
49
Appendix A: Haskell County Recovery Plots
50
A
B
C
Figure A - 1: Horner recovery estimation HS1, (a) 2007-08; and 2008-09: (b) entire
recovery period, (c) only the recovery after the final pumping event.
51
A
HS2 - Horner 07-08
140
y = -160.69Ln(x) + 132.99
120
H(@t'=0) = 2465.51
+ 132.99 =
Recovered: 2598.5
100
80
60
40
Series1
20
Series2
Log. (Series2)
0
1000
100
B
10
1
HS2 - Horner 07-08, Period 2
60
H(@t'=0) = 2531.53
58.14 =
Recovered: 2589.67
50
y = -73.248Ln(x) + 58.138
40
30
20
Series1
10
Series2
Log. (Series2)
0
1000
100
10
1
Figure A - 2: Horner recovery estimation HS2, 2007-08; (A) entire recovery period, (B)
only the recovery after the final pumping event.
52
HS2- Horner 08-09
A
140
H(@t'=0)=2462.80
+123.18 =
Recovered: 2585.98
y = -88.506Ln(x) + 123.18
120
y = -233.99Ln(x) + 137.42
100
H(@t'=0)=2462.80
+137.42 =
Recovered: 2600.22
80
Series1
Series2
60
Series3
40
Log. (Series2)
Log. (Series3)
20
0
1000
100
B
10
1
HS2- Horner 08-09, Period 2
70
H(@t'=0)=2522.04
+63.47 =
Recovered: 2585.51
y = -67.075Ln(x) + 63.468
60
50
H(@t'=0)=2522.04
+60.93 =
Recovered: 2582.97
y = -65.675Ln(x) + 60.928
40
30
Series1
Series2
20
Series3
Log. (Series2)
10
Log. (Series3)
0
1000
100
10
1
Figure A - 3: Horner recovery estimation HS2, 2008-09; (A) entire recovery period, (B)
only the recovery after the final pumping event.
53
HS2 Horner 09-10
120
H(@t'=0) = 2475
+108.23 =
Recovered: 2583.23
100
80
y = -133.14Ln(x) + 108.23
60
40
20
Series1
Series2
0
Log. (Series2)
-20
1000
100
10
Figure A - 4: Horner recovery estimation HS2, 2009-10.
54
1
A
B
C
Figure A - 5: Horner recovery estimation HS3, (a) 2007-08, (b) 2008-09, (c) 2009-10.
55
HS4 - Horner 08-09
90
h(@t'=0) = 2501.92
+80.55=
Recovered: 2582.47
80
y = -91.983Ln(x) + 80.55
70
60
h(@t'=0) = 2501.92
+71.914=
Recovered: 2573.83
y = -73.272Ln(x) + 71.914
50
40
Series1
30
Series2
Series3
20
Log. (Series2)
Log. (Series3)
10
0
1000
100
10
1
Figure A - 6: Horner recovery estimation HS4, 2008-09.
HS4 Horner 09-10
60
H(@t'=0) = 2529.34
+51.33 =
Recovered: 2580.67
50
y = -90.798Ln(x) + 51.329
40
Series1
30
Series2
Log. (Series2)
20
10
0
1000
100
10
Figure A - 7: Horner recovery estimation HS4, 2009-10.
56
1
HS5 - Horner 07-08
10
9
y = -16.284Ln(x) + 9.4011
H(@t'=0) = 2583.47
9.40 =
Recovered: 2592.87
8
7
6
5
Series1
4
Series2
Log. (Series2)
3
2
1
0
1000
100
10
Figure A - 8: Horner recovery estimation HS5, 2007-08.
57
1
HS5 - Horner 08-09
A
8
h(@t'=0) = 2581.99
+7.02=
Recovered: 2589.01
7
6
y = -29.179Ln(x) + 7.0179
5
h(@t'=0) = 2581.99
+4.97=
Recovered: 2586.96
4
y = -8.9567Ln(x) + 4.9734
3
Series1
Series2
Series3
Log. (Series2)
2
1
Log. (Series3)
0
1000
100
B
10
1
HS5 - Horner 08-09
6
H(@t'=0)=2582.98
+5.79=
Recovered: 2588.77
5
y = -4.4208Ln(x) + 5.7919
H(@t'=0)=2582.98
+3.01=
Recovered: 2585.99
4
3
y = -0.3684Ln(x) + 3.0122
2
Series1
Series2
Series3
Log. (Series2)
Log. (Series3)
1
0
10000
1000
100
10
1
Figure A - 9: Horner recovery estimation HS5, 2008-09; (A) entire recovery period, (B)
only the recovery after the final pumping event.
58
HS5 Horner 09-10
4.5
H(@t'=0) =2579.90
+4.29 =
Recovered: 2584.19
4.0
3.5
3.0
y = -16.952Ln(x) + 4.2939
2.5
2.0
1.5
Series1
Series2
1.0
Log. (Series2)
0.5
0.0
1000
100
10
Figure A - 10: Horner recovery estimation HS5, 2009-10.
59
1
HS6 - Horner 07-08
A
9
H(@t'=0)=2581.767
+7.62 =
Recovered: 2589.49
8
7
y = -6.8711Ln(x) + 7.6182
6
5
4
3
2
Series1
Series2
1
Log. (Series2)
0
1000
100
10
1
HS6 - Horner 07-08, Period 2
B
1.6
1.4
H(@t'=0)=2586.20
+1.42 =
Recovered: 2587.62
1.2
1.0
y = -1.8734Ln(x) + 1.4162
0.8
0.6
Series1
Series2
0.4
Log. (Series2)
0.2
0.0
1000
100
10
1
Figure A - 11: Horner recovery estimation HS6, 2007-08; (A) entire recovery period, (B)
only the recovery after the final pumping event.
60
HS6 - Horner 08-09
12
H(@t'=0)=2574.66
+10.07=
Recovered: 2584.73
y = -9.9422Ln(x) + 10.077
H(@t'=0)=2574.66
+8.37=
Recovered: 2583.07
10
y = -3.431Ln(x) + 8.3659
8
6
Series1
4
Series2
Series3
2
Log. (Series2)
Log. (Series3)
0
1000
100
10
1
Figure A - 12: Horner recovery estimation HS6, 2008-09.
HS6 Horner 09-10
8
H(@t'=0) = 2573.64
+6.90 =
Recovered: 2580.54
7
6
y = -11.671Ln(x) + 6.9044
5
4
3
Series1
Series2
2
Log. (Series2)
1
0
1000
100
10
Figure A - 13: Horner recovery estimation HS6, 2009-10.
61
1
HS7 - Horner 07-08
45
H(@t'=0)=2551.91
+42.05 =
Recovered: 2593.96
40
y = -54.422Ln(x) + 42.049
35
30
25
20
15
Series1
10
Series2
Log. (Series2)
5
0
1000
100
10
Figure A - 14: Horner recovery estimation HS7, 2007-08.
62
1
HS7 - Horner 08-09
A
30
H(@t'=0)=2558.28
+28.95=
Recovered: 2587.13
y = -33.443Ln(x) + 28.954
25
H(@t'=0)=2558.28
+26.06=
Recovered: 2584.34
20
y = -38.902Ln(x) + 26.059
15
Series1
Series2
Series3
10
Log. (Series2)
Log. (Series3)
5
0
1000
100
B
10
1
HS7 - Horner 08-09, Period 2
16
H(@t'=0)=2571.88
+15.76=
Recovered: 2587.64
y = -27.655Ln(x) + 15.755
14
12
10
8
6
H(@t'=0)=2571.88
+9.24=
Recovered: 2581.12
Series1
4
Series2
2
Series3
Log. (Series2)
0
Log. (Series3)
y = -1.8074Ln(x) + 9.2439
-2
1000
100
10
1
Figure A - 15: Horner recovery estimation HS7, 2008-09; (A) entire recovery period, (B)
only the recovery after the final pumping event.
63
HS7 Horner 09-10
20
H(@t'=0) = 2565.76
+18.31 =
Recovered: 2584.07
18
16
y = -29.043Ln(x) + 18.306
14
12
10
8
6
Series1
Series2
4
2
Log. (Series2)
0
1000
100
10
Figure A - 16: Horner recovery estimation HS7, 2009-10.
64
1
A
HS8 - Horner 07-08
y = -1.809Ln(x) + 3.9443
4.5
H(@t'=0)=2593.35
+3.94 =
Recovered: 2597.29
4.0
3.5
3.0
2.5
c
2.0
1.5
1.0
Series1
Series2
0.5
Log. (Series2)
0.0
1000
100
10
1
HS8 - Horner 07-08, Period 2
B
4.5
y = -1.0097Ln(x) + 3.618
H(@t'=0)=2593.60
+3.62 =
Recovered: 2597.22
4
3.5
3
2.5
2
1.5
Series1
1
Series2
Log. (Series2)
0.5
0
1000
100
10
1
Figure A - 17: Horner recovery estimation HS8, 2007-08; (A) entire recovery period, (B)
only the recovery after the final pumping event.
65
HS8 - Horner 08-09
5
H(@t'=)0=2588.81
+4.12=
Recovered: 2592.93
y = -2.5693Ln(x) + 4.1243
4
3
2
H(@t'=)0=2588.81
+3.59=
Recovered:2592.4
1
Series1
Series2
y = -0.704Ln(x) + 3.5891
Series3
0
Log. (Series2)
Log. (Series3)
-1
1000
100
10
1
Figure A - 18: Horner recovery estimation HS8, 2008-09.
HS8 Horner 09-10
7
H(@t'=0) = 2583.89
+5.49 =
Recovered: 2583.28
6
5
y = -6.3263Ln(x) + 5.4941
4
3
2
Series1
Series2
1
Log. (Series2)
0
1000
100
10
Figure A - 19: Horner recovery estimation HS8, 2009-10.
66
1
A
B
C
Figure A - 20: Horner recovery estimation HS9, (a) 2007-08, (b) 2008-09, (c) 2009-10.
67
A
B
C
Figure A - 21: Horner recovery estimation HS11, 2008-09; (a) entire recovery period, (b)
only the recovery after the final pumping event, and (c) 2009-10.
68
HS12 - Horner 07-08
3.5
H(@t'=0)=2590.64
+2.38 =
Recovered: 2593.02
3
2.5
y = -5.4926Ln(x) + 2.383
2
Series1
Series2
1.5
Log. (Series2)
1
0.5
0
1000
100
10
Figure A - 22: Horner recovery estimation HS12, 2007-08.
69
1
HS15 - Horner 07-08
2.5
H(@t'=0)=2590.47
+1.56 =
Recovered: 2592.03
2.0
Series1
1.5
y = -3.5489Ln(x) + 1.5645
Series2
Log. (Series2)
1.0
0.5
0.0
-0.5
1000
100
10
1
Figure A - 23: Horner recovery estimation HS15, 2007-08.
HS15 Horner 09-10
3
H(@t'=0) = 2581.58
+1.73 =
Recovered: 2583.31
2
2
Series1
1
Series2
y = -5.2074Ln(x) + 1.725
Log. (Series2)
1
0
-1
1000
100
10
Figure A - 24: Horner recovery estimation HS15, 2009-10.
70
1
HS17 - Horner 07-08
A
1.4
H(@t'=0)=2590.42
+0.26 =
Recovered: 2590.68
1.2
1
Series1
0.8
Series2
Log. (Series2)
0.6
y = -0.1817Ln(x) + 0.2568
0.4
0.2
0
-0.2
1000
100
B
10
1
HS17 - Horner 07-08, Period 2
1.6
H(@t'=0)=2590.31
+1.20 =
Recovered: 2591.51
1.4
1.2
1
Series1
0.8
Series2
Log. (Series2)
0.6
y = -4.3765Ln(x) + 1.1989
0.4
0.2
0
1000
100
10
1
Figure A - 25: Horner recovery estimation HS17, 2007-08; (A) entire recovery period, (B)
only the recovery after the final pumping event.
71
HS18 - Horner 07-08
A
120
h(@t'=0)=2492.91
+105.8 =
Recovered: 2598.71
y = -102.22Ln(x) + 105.8
100
80
Series1
Series2
Log. (Series2)
60
40
20
0
1000
100
B
10
1
HS18 - Horner 07-08, Period 2
50
h(@t'=0) = 2550.07
+43.02 =
Recovered: 2593.09
y = -5.8812Ln(x) + 43.017
45
40
35
30
25
20
15
10
Series1
Series2
5
Log. (Series2)
0
1000
100
10
1
Figure A - 26: Horner recovery estimation HS18, 2007-08; (A) entire recovery period, (B)
only the recovery after the final pumping event.
72
HS18 - Horner 08-09
A
100
H(@t'=0)= 2492.88
+95.32 =
Recovered: 2588.20
y = -56.443Ln(x) + 95.315
90
80
H(@t'=0)= 2492.88
+104.43 =
Recovered: 2597.31
y = -163Ln(x) + 104.43
70
60
50
Series1
40
Series2
30
Series3
20
Log. (Series2)
Log. (Series3)
10
0
1000
100
B
10
1
HS18 - Horner 08-09, Period 2
37
H(@t'=0)= 2550.88
+35.85 =
Recovered: 2586.73
y = -5.3818Ln(x) + 35.849
35
H(@t'=0)= 2550.88
+34.62 =
Recovered: 2585.50
33
y = -0.8101Ln(x) + 34.62
31
29
Series1
Series2
Series3
27
Log. (Series2)
Log. (Series3)
25
1000
100
10
1
Figure A - 27: Horner recovery estimation HS18, 2008-09; (A) entire recovery period, (B)
only the recovery after the final pumping event.
73
HS18 Horner 09-10
100
H(@t'=0) = 2495.88
+87.25 =
Recovered: 2583.13
90
y = -81.76Ln(x) + 87.248
80
70
60
50
40
30
Series1
20
Series2
Log. (Series2)
10
0
1000
100
10
Figure A - 28: Horner recovery estimation HS18, 2009-10.
74
1
HS20 - Horner 07-08
A
18
h(@t'=0) = 2580.67
+16.16 =
Recovered: 2596.83
y = -1.6223Ln(x) + 16.161
16
14
12
10
8
Series1
Series2
6
Log. (Series2)
4
2
0
1000
100
10
1
HS20 - Horner 07-08, Period 2
B
18
y = -0.0364Ln(x) + 15.646
h(@t'=0)=2580.92' AMSL
Horner Recovery = +15.65
Recovered h = 2596.57' AMSL
16
14
12
y = -0.6225Ln(x) + 15.612
10
h(@t'=0)=2580.92' AMSL
Horner Recovery = +15.61
Recovered h = 2596.53' AMSL
8
6
Series1
Series2
4
Log. (Series2)
Log. (Series1)
2
0
1000
100
10
1
Figure A - 29: Horner recovery estimation HS20, 2007-08; (A) entire recovery period, (B)
only the recovery after the final pumping event.
75
HS20 Horner 09-10
A
25
y = -3.8785Ln(x) + 23.244
H(@t'=0) = 2565.47
+23.24 =
Recovered: 2588.71
20
15
Series1
10
Series2
Log. (Series2)
5
0
-5
1000
B
100
10
1
HS20 Horner 09-10, Period 2
25
y = -3.7171Ln(x) + 22.852
H(@t'=0) = 2565.86
+22.852 =
Recovered: 2588.712
20
15
Series1
Series2
10
Log. (Series2)
5
0
1000
100
10
1
Figure A - 30: Horner recovery estimation HS20, 2009-10; (A) entire recovery period, (B)
only the recovery after the final pumping event.
76
HS21 - Horner 08-09
6
H(@t'=0) = 2588.56
+4.35 =
Recovered: 2592.91
5
y = 0.036Ln(x) + 4.3541
H(@t'=0) = 2588.56
+4.70 =
Recovered: 2593.26
4
y = -1.0761Ln(x) + 4.7027
3
Series1
2
Series2
Series3
1
Log. (Series2)
Log. (Series3)
0
1000
100
10
1
Figure A - 31: Horner recovery estimation HS21, 2008-09.
HS21 Horner 09-10
25
H(@t'=0) = 2569.02
+20.12 =
Recovered: 2589.14
y = -1.4469Ln(x) + 20.122
20
15
10
Series1
5
Series2
Log. (Series2)
0
1000
100
10
Figure A - 32: Horner recovery estimation HS21, 2009-10.
77
1
HS28 - Horner 08-09
1.4
1.2
H(@t'=0) = 2582.31
+0.95 =
Recovered: 2583.26
1.0
y = -0.6986Ln(x) + 0.9544
0.8
H(@t'=0) = 2582.31
+0.86 =
Recoverd: 2583.17
y = -0.4073Ln(x) + 0.8594
0.6
0.4
Series1
0.2
Series2
Series3
0.0
Log. (Series2)
Log. (Series3)
-0.2
1000
100
10
Figure A - 33: Horner recovery estimation HS28, 2008-09.
78
1
HS29 - Horner 08-09
4.0
3.5
3.0
H(@t'=0) = 2580.43
+2.85 =
Recovered: 2583.28
2.5
2.0
y = -0.4429Ln(x) + 2.8528
Series1
1.5
Series2
Series3
1.0
Log. (Series2)
0.5
Log. (Series3)
0.0
1000
100
10
Figure A - 34: Horner recovery estimation HS29, 2008-09.
79
1
HS30 - Horner 07-08
A
1.6
h(@t'=0) = 2597.98
+1.64 =
Recovered: 2599.62
1.4
1.2
1
y = -10.323Ln(x) + 1.6411
Series1
0.8
Series2
Log. (Series2)
0.6
0.4
0.2
0
1000
100
10
1
HS30 - Horner 07-08, Period 2
B
0.9
h(@t'=0) = 2598.41
+0.45 =
Recovered: 2598.86
0.8
0.7
Series1
0.6
Series2
y = -0.1658Ln(x) + 0.4494
Log. (Series2)
0.5
0.4
0.3
0.2
0.1
0
1000
100
10
1
Figure A - 35: Horner recovery estimation HS30, 2007-08; (A) entire recovery period, (B)
only the recovery after the final pumping event.
80
A
B
C
Figure A - 36: Horner recovery estimation HS31, (a) 2007-08, (b) 2008-09, and (c) 2009-10.
81
Appendix B: Thomas County Recovery Plots
82
Recovery Pumping
#1
#2
2976
Elevation of Water Level (ft ASL)
2974
2973
37.0
36.0
35.0
May 12, 2975.31'
34.0
2972
33.0
2971
32.0
2970
BP (ft H2O)
2975
Recovery Pumping
Pumped:
#1: 5/12 - 5/16
#2: 5/18 - 5/21
h@t':
#1: 2973.33'
#2: 2973.65'
Recovery:
#1: 34h
#2: 377h
31.0
Sept.8, 2970.53'
2968
29.0
ar
-0
8
7
Corrected WL
P2-ET-80hrs
Corrected WL
P2-ET
Early Time - 80h
P2-LT
31
-M
1M
1D
31
-A
ug
ec
-0
-0
7
-0
1Ju
n
Transducer WL
P1-ET
ay
-0
8
30.0
7
2969
Early Time
Colby BP
Late Time
Site BP
Figure B - 1: Thomas Co. index well hydrograph, barometric pressure, and corrected
water level, 2007-08 recovery.
Single Well (SW, tp=5d)
SW, Early Time 80hrs
SW, Early Time
SW, Late Time
Season Pumping (SP, tp = 60d?)
SP, Early Time 80hrs
SP, Early Time
SP, Late Time
Log. (SP, Late Time)
Log. (SW, Late Time)
Expon. (SP, Early Time 80hrs)
Log. (SW, Early Time)
Log. (SP, Early Time)
Expon. (SW, Early Time 80hrs)
07-08 Recovery, TH Index Well
6.0
y = -5.4455Ln(x) + 6.0008
5.0
y = 1.1093e
3.0
y = -46.486Ln(x) + 5.714
Recovery Level
4.0
Initial Water Level Elevation: 2970.53'
Predicted Recovery Water-Level Elevations:
Early Time Data:
SP, 80h = 2971.64' (+1.11')
SP (ignoring earliest data) = 2971.75' (+1.22')
SW, 80h = 2971.64' (+1.11')
SW (ignoring earliest data) = 2971.69' (+1.16')
Late Time Data:
SP = 2976.53' (+6.00')
SW = 2976.24' (+5.71')
Maximum Observed Water-Level: 2975.54'
2.0
y = -0.0367Ln(x) + 1.223
1.0
-0.0005x
y = 1.1153e
-0.0059x
y = -0.0457Ln(x) + 1.1558
0.0
10000
1000
*Note: Unknown season pumping; assumed @ 60d
100
10
1
(tp+t')/t'
Figure B - 2: Horner recovery estimations, Thomas Co. index well, 2007-08 recovery
season.
83
07-08 Recovery, Late Recovery Season Pumping
3.00
Predicted Recovery
Water-Level Elevations:
Early Time Data:
1st pumping period (34h only)
tp = 5d (SW1): 2974.63'
tp = 4.5d (SP1): 2974.63'
2nd pumping period (1st 80 hrs):
tp = 5d (SW2): 2974.96'
tp = 2.83d (SP2): 2974.96'
Early Time Data (ignoring earliest):
SP1: 2975.33' (+2.00')
SW1: 2975.35'
SP2: 2975.18'
SW2: 2975.23'
Late Time Data:
SP1 & SW1 = no data
SW2 = 2976.30'
SP2 = 2976.23'
Initial Water Level Elevations:
1st pumping period = 2973.33'
2nd pumping period = 2973.65'
y = -3.2893Ln(x) + 2.6496
2.00
y = -0.4263Ln(x) + 2.02
1.50
y = -0.4348Ln(x) + 1.9973
y = -5.0633Ln(x) + 2.5773 1.00
Recovery Level (ft)
2.50
y = -0.2149Ln(x) + 1.5802
y = -0.2489Ln(x) + 1.5321
0.50
-0.01x
y = 1.3138e
-0.0099x
y = 1.3189e
-0.009x
y = 1.3124e
-0.0175x
y = 1.3289e
1000.00
100.00
SW-Period 1
SP-Period 2
SW-P2-Late Time
SW-P2-Early Time
Expon. (SW-Period 1)
Expon. (SP-P2-Early Time - 80hrs)
Log. (SP-P2 Early Time)
Log. (SP-P1-Early Time)
(tp+t')/t'
10.00
SP-Period 1
SW-P2-Early Time - 80 hrs
SP-P2-Late Time
SP-P1-Early Time
Expon. (SP-Period 1)
Log. (SW-P2-Late Time)
Log. (SW-P2-Early Time)
0.00
1.00
SW-Period 2
SP-P2-Early Time - 80hrs
SP-P2 Early Time
SW-P1-Early Time
Expon. (SW-P2-Early Time - 80 hrs)
Log. (SP-P2-Late Time)
Log. (SW-P1-Early Time)
Figure B - 3: Horner recovery estimations, Thomas Co. index well, following pumping
periods #1 and #2 in 2008.
84
2008-09 Recovery, Thomas Index Well
Recovery Pumping
#1
#2 #3
37.0
2975
36.0
2974
35.0
2973
34.0
6/2 - 2973.78'
5/26 - 2973.08'
3/23 - 2972.95'
2972
Recovery Pumping
Pumped:
Recovery:
#1: 7.54d
1394h
#2: 5.96d
95h
#3: 2.67d
532h
2971
9/14 - 2970.15'
2970
33.0
32.0
31.0
2968
29.0
Corrected WL
P1-LT
Site BP
27
-M
ar
26
-J
un
-0
9
08
26
-D
ec
-
26
-S
ep
-0
8
27
-M
ar
26
-J
un
-0
8
Corrected WL
P1-ET
Colby BP
-0
9
30.0
-0
8
2969
Transducer WL
LT
P3-LT
BP (ft H2O)
Elevation of Water Level (ft ASL)
2976
ET 53h
P2-ET
ET
P3-ET
Figure B - 4: Thomas Co. index well hydrograph, barometric pressure, and corrected
water level, 2008-09 recovery.
2008-09 TH Co. Index Well Recovery
7.0
Initial Water Level Elevation: 2970.15'
Predicted Recovery Water-Level Elevations:
Early Time Data:
SP (tp=2832h), 1st 53h of recovery = 2971.33' (+1.78')
SP (ignoring earliest data) = 2971.48' (+1.33')
SW (tp=5h), 1st 53h of recovery = 2971.33' (+1.18')
SW (ignoring earliest data) = 2971.36' (+1.21')
Late Time Data:
SP = 2976.41' (+6.26')
SW = 2975.87' (+5.72')
Maximum Observed Water-Level: 2975.09'
6.0
y = -3.95Ln(x) + 6.26
5.0
4.0
3.0
y = -53.174Ln(x) + 5.7181
Recovery Level (ft)
Season Pumping (SP, tp= 118d)
Single Well (SW, tp=5d)
SP, Early Time
SP, Early Time, "Flat" Section
SP, Late Time
SW, Early Time
SW, Early Time, "Flat"
SW, Late Time
Expon. (SP, Early Time)
Log. (SP, Early Time, "Flat" Section)
Log. (SP, Late Time)
Expon. (SW, Early Time)
Log. (SW, Early Time, "Flat")
Log. (SW, Late Time)
2.0
-0.0004x
y = 1.1783e
y = -0.0459Ln(x) + 1.3323
1.0
y = -0.0552Ln(x) + 1.2126
-0.0091x
y = 1.1885e
0.0
10000
1000
100
10
1
(Tp+T')/T'
Figure B - 5: Horner recovery estimations, Thomas Co. index well, 2008-09 recovery
season.
85
Thomas Co. 2008-09 Recovery, Late Recovery Pumping Periods
2.5
y = -2.74Ln(x) + 2.30
2.0
1.5
y = -1.20Ln(x) + 1.50
1.0
Recovery Level (ft)
Initial Water Level Elevations:
Predicted Recovery
1st pumping period = 2972.96'
Water-Level Elevations:
2nd pumping period = 2973.08'
Early Time Data (1st 50h):
3rd pumping period = 2973.71'
1st pumping period = 2974.13' (+1.17')
2nd pumping period = 2974.62' (+1.54')
3rd pumping period = 2974.74' (+1.03')
Early Time Data (flat):
y = -0.12Ln(x) + 1.73
1st pumping period = 2974.39' (+1.43')
2nd pumping period = 2974.81' (+1.73')
3rd pumping period = 2974.75' (+1.04')
Late Time Data:
y = -0.15Ln(x) + 1.43
1st pumping period = 2975.26' (+2.30')
2nd pumping period = Insuff. Data
3rd pumping period = 2975.21 ' (+1.50')
y = -0.11Ln(x) + 1.04
-0.006x
y = 1.5534e
0.5
-0.006x
y = 1.1773e
-0.0372x
y = 1.069e
0.0
1000
100
1st pumping period
2nd, ET all
3rd, Early Time
Log. (1st, Early Time)
Expon. (2nd, ET all)
2nd pumping period
3rd, ET all
1st, Late Time
Log. (2nd, Early Time)
Expon. (1st, ET all)
10
(tp+t')/t'
3rd pumping period
1st, Early Time
3rd, Late Time
Log. (3rd, Early Time)
Expon. (3rd, ET all)
1
1st, ET all
2nd, Early Time
Log. (3rd, Late Time)
Log. (1st, Late Time)
Figure B - 6: Horner recovery estimations, Thomas Co. index well, following pumping
periods #1, #2, and #3 in 2009.
86
09-10 Recovery Season, Thomas Co. Index Well
2977
June 4
Transducer WL
37.0
Corrected WL
2976
Early Recovery
June 24
Late Recovery B
2975
Late Recovery B2
35.0
34.0
2973
33.0
2972
BP (ft H2O)
Site BP
2974
32.0
Aug. 27, 2971.11'
2970
30.0
2969
29.0
0
-S
un
26
27
-J
ar
-M
ep
-1
0
-1
-0
9
ec
28
26
-D
ep
-S
-J
27
26
9
-0
un
-0
ar
-M
28
-1
0
31.0
-0
9
2971
9
Elevation of Water Level (ft ASL)
36.0
Late Recovery A
July 27
Figure B - 7: Thomas Co. index well hydrograph, barometric pressure and corrected water
level, 2009-10 recovery.
09-10 Recovery, Th Co. Index Well
9.00
Initial Water Level Elevation: 2971.11'
Predicted Recovery Water-Level Elevations:
Early Time Data:
SP (tp=1512h), 1st 60h of recovery = 2972.52' (+1.41')
SP (ignoring earliest data) = 273.02' (+1.91')
SW (tp=5h), 1st 60h of recovery = 2972.52' (+1.41')
SW (ignoring earliest data) = 2972.71' (+1.60')
Late Time Data:
SP = 2977.82' (+6.71')
SP2 = 2978.14' (+7.03')
SW = 2976.66' (+5.55')
Maximum Observed Water-Level: 2976.20'
8.00
y = -10.282Ln(x) + 7.0302
7.00
y = -8.851Ln(x) + 6.7055
6.00
Recovery Level (ft)
Season Pumping (SP- tp=63d)
Single Well (SW-tp=5d)
SW, early recovery
SP, Early recovery
SP, late recovery
SW, late recovery
SP, late recovery 2
SP (1st 60 hrs)
SW (1st 60 hrs)
Log. (SW, early recovery)
Log. (SP, Early recovery)
Log. (SW, late recovery)
Log. (SP, late recovery)
Log. (SP, late recovery 2)
Expon. (SP (1st 60 hrs))
Expon. (SW (1st 60 hrs))
5.00
y = -49.069Ln(x) + 5.5532
4.00
3.00
2.00
y = -0.1507Ln(x) + 1.9093
y = -0.1733Ln(x) + 1.5988 1.00
-0.0011x
-0.0134x
y = 1.4069e
10000.00
y = 1.4243e
1000.00
100.00
10.00
0.00
1.00
(tp+t')/t'
Figure B - 8: Horner recovery estimations, Thomas Co. index well, 2009-10 recovery
season.
87
10-11 Recovery, TH Co. Index Well
2977
37.0
Transducer WL
Corrected WL
2976
36.0
Early Time
SP Late Time
35.0
2974
34.0
2973
33.0
2972
32.0
9/6 2971.08'
2971
31.0
-1
0
-1
0
ec
ov
26
-D
-N
-O
25
26
26
26
ep
-S
ug
-A
-J
27
ct
-1
-1
0
-1
ul
-1
0
26
-J
un
ay
27
-M
-A
26
-1
0
pr
-1
-1
ar
-M
0
29.0
0
2969
0
30.0
-1
0
2970
27
BP (ft H2O)
Site BP
0
Elevation of Water Level (ft ASL)
SW Late Time
2975
Figure B - 9: Thomas Co. index well hydrograph and corrected water level, 2010-11
recovery.
Thomas Co. 2010-11 Recovery
3.00
Initial Elevation: 2971.08'
SP Early Time - all: 2972.38' (+1.30')
SW Early Time - all: 2972.38' (+1.30')
SP Early Time: 2972.59' (+1.51')
SW Early Time: 2972.43' (+1.35')
SP Late Time: 2973.81' (+2.73')
SW Late Time: 2973.34' (+2.26')
Max Observed (Nov. 2010): 2973.30'
2.50
2.00
y = -0.48Ln(x) + 2.73
y = -1.38Ln(x) + 2.26
1.50
y = -0.07Ln(x) + 1.51
y = -0.08Ln(x) + 1.35
Recovery Level (ft)
Single Well (SW, tp=5d)
Season Pumping (SP, tp=77d)
SP, Early Time all
SW, Early Time all
SP, Early Time
SW, Early Time
SP, Late Time
SW, Late Time
Log. (SW, Late Time)
Log. (SP, Early Time)
Log. (SW, Early Time)
Log. (SP, Late Time)
Expon. (SP, Early Time all)
Expon. (SW, Early Time all)
1.00
0.50
-0.0008x
-0.0117x
y = 1.3014e
10000.00
y = 1.3158e
1000.00
100.00
10.00
0.00
1.00
(tp+t')/t'
Figure B - 10: Horner recovery estimations, Thomas Co. index well, 2010-11 recovery
season.
88
B.
40
TH3 09-10
100
90
Recovery Level (ft)
Water Level (ft) Above Sensor
A. 110
80
70
60
50
40
30
20
10
30
20
Sensor Malfunction
0
10
0
1
10
100
1000
10000
Recovery Time (hrs)
8/1/09
40
12/1/09
4/1/10
8/1/10
D.
30
Recovery Level (ft)
From Minimum Water Level
Recovery Level (ft)
From Minimum Water Level
C.
20
10
60
40
20
0
0
1000
100
10
1
2
Normalized Recovery Time [(tp+t')/t']
Normalized Recovery Time [(tp+t')/t']
1
Figure B - 11: 2009-10 hydrograph (A) and recovery from Thomas Co. well TH3.
Recovery is plotted as semi-log recovery (B) and as Horner recovery (C) and (D). The
reference time and water level elevation for water level recovery are 8/25/2009 10:00,
2948.92 ft..
89
A.13.6
1.6
B.
Recovery Level (ft)
From Minimum Water Level
Water Level (ft) Above Sensor
Th10 09-10
13.2
12.8
12.4
1.2
0.8
0.4
0
12
-0.4
11.6
8/1/09
1.6
12/1/09
4/1/10
10
100
1000
10000
Recovery Time (hrs)
Since Minimum Water Level
8/1/10
D.1.6
1.2
Recovery Level (ft)
From Minimum Water Level
Recovery Level (ft)
From Minimum Water Level
C.
1
0.8
0.4
0
1.2
0.8
0.4
0
-0.4
-0.4
1000
100
10
1
2 Normalized Recovery Time [(tp+t')/t']
Normalized Recovery Time [(tp+t')/t']
Since Minimum Water Level
Since Minimum Water Level
1
Figure B - 12: 2009-10 hydrograph (A) and recovery from Thomas Co. well TH10.
Recovery is plotted as semi-log recovery (B) and as Horner recovery (C) and (D). The
reference time for water level recovery is 8/25/2009 10:00. Surface elevation of the well is
unknown, but estimated at 3132 ft AMSL from the Google Earth digital elevation model.
90
Appendix C: Using the KGS Barometric Pressure Correction Spreadsheet and
Related Software (KGS_BRF.xls and kgs_brf.exe)
Introduction
Changes in barometric pressure affect water levels in all three index wells. These barometricpressure-induced fluctuations in water level can introduce uncertainty into estimates of annual
water-level changes at the index wells and elsewhere. The KGS has therefore developed an
Excel spreadsheet to remove the effect of barometric-pressure fluctuations from water-level
measurements. This spreadsheet calculates a Barometric Response Function (BRF) to
characterize the relationship between changes in barometric pressure and changes in water level.
This BRF is then used to remove (correct) the impact of fluctuations in barometric pressure from
the water-level measurements. Further information about BRFs is provided in Appendix E.
File Management
The KGS barometric pressure correction software has two components, an Excel worksheet
contained in the workbook KGS_BRF.xls and a compiled program (executable) named
kgs_brf.exe. The Excel worksheet serves as a front end to the executable, providing a template
for managing the water level, barometric pressure, and (optionally) earth tide data. The
worksheet contains three buttons, one to fill gaps in the data records, one to run the computations
for estimating a BRF and also to correct water levels using that BRF, and one to correct water
levels using a BRF that has already been computed. The Visual Basic code that is behind these
latter two buttons reads information from the worksheet, writes it out to a set of input files for the
executable, runs the executable, and then reads the output from the executable back into Excel.
This means that the Excel spreadsheet cannot work without access to the executable. At the
moment, this means that a copy of the executable file, kgs_brf.exe, has to exist in the folder that
contains the Excel workbook with which you are working.
You may make copies of kgs_brf.exe using any of the methods provided by Windows Explorer
– selecting an existing copy of the file, then copying and pasting the new copy in the desired
folder, selecting and ctrl-dragging, etc. To see the full file name, with the extension, you will
need to tell Windows Explorer to show you file extensions. But even if you don’t, the Excel file,
KGS_BRF.xls, should be tagged with an Excel icon, distinguishing it from the executable.
Furthermore, it is quite likely that you will end up using workbooks that are named something
other than KGS_BRF.xls, anyway. The Excel Visual Basic code is directly attached to the
Input_Template worksheet in the KGS_BRF.xls. This means that you can make copies of this
worksheet and/or workbook, using any name you please, and the code will be part of each new
copy. This allows you to create and save copies of the Input_Template worksheet using more
meaningful names without “breaking” the software. But, again, you will need to copy the
executable, kgs_brf.exe, to each folder that you work in. You are not allowed to change the
name of kgs_brf.exe because the Excel VB code looks for it by that name.
91
The executable program has been designed so that it can be used on its own, without the Excel
front end. Using it involves creating a set of plain text input files (a parameter file and input data
files) and then running the program in a DOS command window. The details of this process will
be explained in another report. The Visual Basic code attached to the Input_Template
worksheet automates the process of generating the input files and reading the output files.
The Excel workbook (and included Visual Basic code) has been created in Excel 2003. It should
also work in more recent versions of Excel.
Macro Security
To be able to run the Visual Basic code included in KGS_BRF.xls, you may need to alter
Excel’s macro security level from its current setting. In Excel 2003, you set the macro security
level by selecting Options… from the Tools menu, then selecting the Security tab on the
Options dialog box, and then clicking the Macro Security… button on that tab. On the
resulting dialog box, you should set the security level to Medium:
92
With the macro security level set to Medium, you will be presented with the following dialog
box when you open KGS_BRF.xls (or any other workbook containing macros):
You should click the Enable Macros button on this dialog box. If you set the macro security
level to Low, then Excel will just open a macro-bearing workbook with the macros enabled,
without asking for your permission. As noted on the Security dialog box, this is not advisable.
93
In Excel 2007, you change the security settings as follows:
Select the Office button in the upper left hand corner of the Excel window to get the Office dropdown menu:
Select the Excel Options button on this menu, then select Trust Center in the list on left side of
the Excel Options dialog box, then click the Trust Center Settings… button (on the lower-ish
right), then select Macro Settings from the list on the left of the Trust Center dialog box, and
then select Enable all macros:
94
This is the same as the Low security setting in Excel 2003. As the dialog box says, this setting is
not recommended, but it will have to do until we figure out how to create digitally signed
macros. Unfortunately, Excel 2007 does not have a macro security level corresponding to the
Medium setting in Excel 2003.
The Input_Template spreadsheet
The (upper left corner of the) Input_Template spreadsheet looks like this:
To use it, you do what the note in Cell A1 says: Copy your data into the template and then press
the Compute BRF or Correct WL button (the latter requires that you have already done the
95
former). This means that you paste your measurement time, water level, and barometric pressure
data into columns A-C, starting at row 20, update the information in the yellow cells
appropriately, and then press the appropriate button. Neither the BRF nor water level correction
computations allow missing values in the measurements. If you have gaps in the data, like the
water level measurements that are missing from cells B24 and B25 above, you should fill them
using the Fill Gaps button, as explained below.
Important: The Visual Basic code looks for each piece of information by cell address. This
means . . . don’t move anything. Just revise the information in place.
In order to avoid mixing up your new data with the data that are already in the worksheet, it is
advisable to delete the old data first, by selecting the data from row 20 on down and then
deleting it. Clearing the cells using the Delete button should be sufficient, or you can really mop
things up by selecting all the cells (or rows) and then selecting Delete… from the Edit menu. If
the new data record is as long or longer than the old data record, so that pasting in the new data
will completely overwrite the old data, then the deletion step is not necessary. However, it is
advisable to delete the old data first, just to be sure.
The code determines the length of the data record based on the measurement time data starting in
cell A20. It reads down this column from row 20 until it finds a blank cell. The cell above this
first blank cell is the last data point in the record, even if there are additional data below the
blank cell.
The measurement times listed in column A do not actually matter to the BRF and water level
correction (WLC) computations. They are solely for informational and plotting purposes. The
BRF and WLC computations assume that the data are (strictly) regularly sampled, with the
sample interval given in cell B9. Time in these computations is given by the sample interval
times the sample number (index). The code behind the Fill Gaps button, however, actually
employs the measurement times and requires that they be in strictly increasing order (each time
is strictly greater than the previous time).
You should modify cells B4-B16 (labels in cells A4-A16) to specify the following information:
Comment (cell B4): This is a note to yourself regarding the data and/or analysis. It will be
passed on to the output BRF and water level correction spreadsheets.
Well (cell B5): The well name
Water Level Units (cell B6): The units of the water level measurements. This cell is
implemented as a pick list allowing selection from the units listed in cells M5-M6 (feet and
meters). See information about units on page 10.
Barometric Pressure Units (cell B7): The units of the barometric pressure measurements. This
cell is implemented as a pick list allowing selection from the units listed in cells P5-P10. See
information about units on page 10.
96
Earth Tide Units (cell B8): The units of the earth tide values. This information is not used if
the number of earth tide lags is set to -1. If earth tide data are employed, the code will accept
any units that you type into cell B8 and the earth tide response coefficients will end up having
units of feet per earth tide unit, whatever that unit may be.
Sample Interval (cell B9): The sample interval for the measurements. The BRF and water
level correction computations assume that the measurements are regularly sampled at the sample
interval, and ignore the actual measurement time values listed in column A (except for the sake
of selecting the data subsets to use for BRF and WLC computations, as described below).
Assuming that these measurement time values are Excel date/time values, then a convenient way
to specify the sample interval is to set cell B9 equal to the difference between the first two
measurement times, that is, cell A21 minus cell A20. This difference will yield a numeric value,
which is in days (e.g., 0.04167 days if the measurements are one hour apart).
Sample Interval Units (cell B10): This entry defines the units of time. If the sample interval is
specified as described above (difference between cells A21 and A20, with those cells containing
Excel date/time values), then the sample interval will be in days.
Number of BP Lags (cell B11): The number of lagged values of barometric pressure to use in
the analysis. This means the number of values preceding the current water level measurement.
A lag of zero means the barometric pressure measurement at the same time as the current water
level measurement, so the number of BP values used in the analysis is actually the number of BP
lags plus 1. You could set the number of BP lags to 0 to use just the zero-lag BP value –
meaning there would still be something to compute. To exclude BP values from the analysis,
you should set the number of BP lags to -1. You would do this only if you wanted to analyze
responses to earth tides alone, but since the code does not yet handle earth tides, this option
presently does not make any sense.
Number of ET Lags (cell B12): Same as above, except for earth tide (ET) values, instead of BP
values. If the number of ET lags is set to -1, then ET values (column D) are not required and
will be ignored if they are present.
BRF Start Date and BRF End Date (cells B13 and B14): The BRF will be computed based on
a subset of the data measured between the two date/time values specified in cells B13 and B14.
The selection includes these two end points, assuming they correspond to actual measurement
times in the data record.
Correction Start Date and Correction End Date (cells B15 and B16): The water level
correction process will be applied to the subset of data between the two date/time values
specified in cells B15 and B16, again including the end points.
Filling Data Gaps
The BRF and water level correction computations do not allow missing values of WL or BP
within the range of measurement times spanned by the BRF or correction start and end dates
97
(cells B13 and B14 or cells B15 and B16). The same applies to ET values when ET is used. For
the sake of illustration, the WL and BP columns shown in the screen shot on page 5 include a
few missing values. You can use the Fill Gaps button to interpolate across gaps within the data
series, like the gap in the water level series represented by the empty cells B24-B25. However,
the Fill Gaps code will not fill empty cells at the beginning or end of the record, like the three
missing BP values represented by cells C20-C22, since this would involve extrapolating beyond
the available data.
The Fill Gaps code performs a linear interpolation between the observed data values on either
side of the gap, interpolating to the provided measurement times for the missing data values.
This code requires that the measurement times be in strictly increasing order and will display an
error message and stop if they are not. Once it is done running, the code will present a dialog
box showing the number of missing data values that it filled in:
As stated by the dialog box, the interpolated values will be highlighted in red:
The red highlighting is a change to the formatting of the cells and will not go away unless you
change the formatting by some mechanism, such as explicitly changing the format or pasting in
new values with formats included. However, the Fill Gaps code will also set (or re-set) the font
color for non-empty cells to black. The reasoning for this behavior is that if we pasted in a new
data record and then ran Fill Gaps, the black and red font colors would then correctly indicate
the measured and interpolated values in this new record, even if we hadn’t bothered to undo the
red formatting of the interpolated cells in the previous record. However, a side effect of this
behavior is that the code also eliminates the highlighting of interpolated cells if we run it again
on a record that contains interpolated values. That is, if we ran Fill Gaps again with the
spreadsheet in the state shown above, then the two interpolated WL values would be taken as
“present” (not missing) and their font would be set to black. The resulting dialog box would also
indicate that the code had filled in 0 WL values. That is, running Fill Gaps more than once on
the same data record will obliterate the distinction between measured and interpolated values.
98
Computing a BRF and Correcting Water Levels
When you have your data in place and have modified the informational (yellow) cells
appropriately, click on the Compute BRF (and Correct WL) button to
1) compute a BRF based on the WL and BP measurements in the spreadsheet with
measurement times between the BRF Data Start and BRF Data End date/times
(inclusive) specified in cells B13 and B14, and
2) use that BRF to correct for the influence of BP variations on the WL measurements in the
spreadsheet with measurement times between the Correction Data Start and Correction
Data End date/times (inclusive) specified in cells B15 and B16.
The coefficients of the computed BRF, along with confidence intervals on those coefficients,
will be written out to a new spreadsheet which is added to the current workbook. The name of
this new spreadsheet will be BRF n, where n is an integer. The code will count all the
spreadsheets in the active workbook whose names start with “BRF” and then set n to that number
plus 1. The code will also add a plot to the BRF worksheet showing the cumulative coefficients
(big A) with error bars.
If ET values are also employed, then the BRF worksheet will also contain the earth tide response
function (ETF) coefficients and a plot of cumulative earth tide coefficients (big B) with the
corresponding error bars.
This new BRF worksheet is yours to do with what you will: rename it, move or copy it, etc. It
contains no links (via formulas) to the original data sheet or to the Visual Basic code and will not
“break” if you move it. Nor does the BRF worksheet contain any VB code of its own, so if you
copied or moved it to a new workbook, you would not be adding any macros to that workbook
(leading to a need to enable macros when you open that workbook). All the VB code is
associated only with the Input_Template worksheet (or copies thereof). However, if you want to
use the BRF contained in this worksheet later to correct other water levels, then you should not
alter the contents of this worksheet. When you correct water levels using a previously calculated
BRF, the water level correction code will expect to find the right information in appropriate cells
in the BRF worksheet.
The corrected water levels will also be written out to a new worksheet, which will be named
WLC n, where n is 1 plus the number of worksheets in the current workbook whose names start
with “WLC”. This worksheet will include a plot showing the original and correct water levels,
along with the barometric pressure values (on the secondary Y axis). This corrected water levels
worksheet is also yours to do with what you will. Unlike the BRF worksheet, there is no need to
be concerned about altering the contents of the WLC worksheet, since it will not be accessed
again by the VB code.
99
The listing of corrected water level values will not start until the number of measurements is
equal to the number of BP lags plus 1. This is because this number of previous BP values has to
be accumulated before the correction can be applied.
Correcting Water Levels (with selected BRF)
It is possible that you will want to correct a series of water level measurements using a BRF
computed using some other series of measurements. You can accomplish this using the Correct
WL (with Selected BRF) button. The correction will be applied to the measurements in the
Input_Template worksheet (or copy thereof), but the BRF coefficients will be read from the
worksheet whose name appears in cell J14 (following the Selected BRF label). Whenever you
compute a new BRF, the code will put the name of the newly generated BRF worksheet into cell
J14 on the Input_Template worksheet. However, you can replace this with the name of any other
BRF worksheet by typing the name of that worksheet into cell J14. The BRF worksheet needs to
reside in the active workbook, but this could be accomplished by copying the BRF worksheet
from some other workbook.
100
Water Level and Barometric Pressure Units
The cells for specifying the measurement units of WL and BP, cells B6 and B7 of the
Input_Template worksheet, are implemented as drop-down pick lists using Excel’s Validation…
option (on the Data menu, at least in Excel 2003). Currently, the list of WL units in cell B6
comes from cells M5 and M6, which contain “feet” and “meters”. Cells N5 and N6 contain the
multipliers needed to convert each of these units to feet, namely 1 and 3.281. The code will use
the multiplier corresponding to the selected units to convert water levels to feet. Similarly, the
allowed BP units are listed in cells P5 to P10, with the multipliers required to convert them to
equivalent feet of water listed in cells Q5 to Q10. The code will use the appropriate multiplier to
convert BP to feet of water:
Additional options could be added to these lists by adding the label for the units to the list in
column M or P and adding the multiplier for conversion to feet to the adjacent cell in column N
or Q. To add the new units to the drop-down list of options (in Excel 2003), select either cell B6
or B7, then select Validation… from the Data menu and expand the list of cells serving as the
Source for the list. For example, to add Atmospheres to the list of allowable BP units, you could
type atm in cell P11 and 33.96 in cell Q11 (one atmosphere corresponds to 33.96 feet of water at
68 degrees F), and then use the Data Validation dialog box to change the Source for the list in
cell B7 to include cell P11:
101
102
Appendix D: KGS Four-township Thomas County Region Water Budget Study
Introduction
In 2005 a group of water right holders in southern Thomas County entered into discussions about
the possibility of voluntarily forming a special groundwater management subunit within
Groundwater Management District Number 4. The area in question was within a candidate
region for designation as a priority subunit, as defined by the Kansas Water Office.
At the request of GMD4, the Kansas Geological Survey undertook a study of the area in order to
assemble and interpret the available hydrogeologic information within the area of interest. This
was formulated as a water budget for the area in question, in order to provide the interested
parties with the best available quantitative estimates so that they could explore possible “what-if”
effects of various decisions or management scenarios.
The study was completed under time pressures imposed by meetings and practical deadlines
already scheduled within GMD4. The attached material was prepared and made available for
both internal and external review on January 12, 2006, and was presented at a public meeting in
GMD4 on January 20.
Although the data assembly and analyses were rigorously and carefully performed, time did not
allow development of the presentation into either a fully technical report or a document
completely oriented to the lay public. In spite of its technical merit, it was not issued as a formal
KGS publication or open-file report because of the lack of stylistic development and
completeness.
The existence of the budget study was one of the reasons for siting the GMD4 experimental
index well within the area. The increasing inventory of quantitative data obtained from that well
and other expansion wells within the study area are in turn meshing with the budget study to
suggest further explanations and hypotheses for characteristics and behavior of the groundwater
resources in the area.
In order to make the findings available in a more formal, citable fashion, the original report is
included as an Appendix in this open-file report.
103
Water Budgets, Four-township Thomas County Region
R. W. Buddemeier, D. P. Young, B. B. Wilson
Background
The four townships outlined in Figure D - 1 are the target of a water budgeting exercise.
Groundwater flow in the area is generally from west to east; this, plus the absence of significant
development to the west and the south make the area of interest rather hydrologically isolated.
There may be some interactions with wells external to the area along the north boundary in
general, and along the south boundary of 9-32, but it seems like a very good first approximation
to treat the region as an isolated entity.
9S 32W
9S 34W
9S 33W
10S 33W
Figure D - 1: T. 9 S. Rgs. 32-34, and T. 10 S., R. 33 W., in Thomas County, with
surrounding area. Dots are water rights locations and crosses indicate monitoring wells.
104
Within the four townships, the wells can be grouped by township or by some other affinity
grouping (for example, N and S of the South Fork Solomon River).
The objectives in constructing the water budget(s) are to obtain better information on the
feasibility and potential effects of instituting a water conservation program in the area that would
extend the usable life of the aquifer, and to provide both general and specific information to the
irrigators in the area.
The conceptual basis is shown in the Figure D - 2. The primary measure of concern is the
amount of groundwater in storage (saturated thickness times area times specific yield). If inputs
are equal to outputs, the water level will remain constant; if not, the volume will change
(reflected in a rise or decline in the water table).
Water Table
Figure D - 2: Terms in the water budget of a region. If water loss is greater than gain, the
elevation of the water table will decline. Typically, we have measurement-based estimates
of annual pumping, precipitation, and annual water table elevation, plus some data on
bedrock elevation, and the specific yield and hydraulic conductivity of the aquifer.
105
Data Used
Data used for the analysis were primarily derived from the KGS section-level database for the
High Plains aquifer (http://hercules.kgs.ku.edu/geohydro/section_data/hp_step1.cfm) and from
updates to that dataset (all section-scale values assigned to section center coordinates prepared
by Brownie Wilson). These updates include
1. Annual water table elevations, 1996 through 2005 (individual year data, not multi-year
averages).
2. Changes in water table elevation, 96-97, 97-98, 98-99, 99-00, 00-01, 01-02, 02-03, 03-04,
and 04-05. These were calculated from the changes in the individual year elevations at
each monitoring well, and the change values interpolated to the section centers.
3. Reported water use for each section for the years 1996 through 2004.
4. Use-density, 2 mile radius - this smoothes the water use by averaging the individual
points over a 2 mile radius, combining and extending the effects of unevenly spaced
wells to give a better picture of the effects on the water table over a reasonable zone of
pumping influence. Only a two-mile radius (rather than the usual 2, 5 and 10 mile
calculations done for the aquifer as a whole) was used in order to minimize the edge
effects that would be substantial because of the extensive and rather distinct boundaries
between irrigated and non-irrigated areas.
5. Hydrographs and measurements for the monitoring wells and others in the area from the
KGS Wizard database (http://www.kgs.ku.edu/Magellan/WaterLevels/index.html).
In addition, we acquired the monthly NCDC precipitation data for Colby, Mingo and Oakley
(http://lwf.ncdc.noaa.gov/oa/ncdc.html). The experience and local knowledge of GMD4 and
DWR staff familiar with the area, and of local irrigators, was also taken into consideration. Other
information available includes the KGS WWC5 well log database and available literature, in
particular the dissertation of Gary Hecox, who performed a detailed model analysis of the GMD4
region (Hecox 2003).
Budget Components – Description and Assessment:
Water Elevations and Changes
Water elevations and changes are measured annually in early January by KGS and DWR. The
primary measurement is of depth to water from a datum, and the best available estimate of the
elevation of the datum is used to calculate the water table elevation at that point. Water table
elevations at other locations are calculated using a triangulation interpolation network (TIN) to
create a calculated surface connecting three wells, and then sampling the elevation of the surface
at the point of interest. Figure D - 3 shows the monitoring wells and TIN network for the region
of interest.
106
Figure D - 3: The four townships of interest are outlined in red; wells used by the annual
monitoring program are indicated by crosses, and points of diversion by dots. The lines
connecting the monitoring wells define the TIN boundaries used to calculate water levels
that fall between the measuring points.
The well network was originally designed based on a well density that corresponded
approximately to one per township, but in some areas this has been highly modified due to
problems in finding accessible wells in good condition.
As a general rule, the results of the annual monitoring program are regarded as being suitable for
assessing changes in the state of the groundwater resource over times of 5-10 years and spatial
scales greater than a township in size. This is because there are a number of potential errors and
uncertainties in assuming that the measured water level is an accurate representation of the
region around it, and all wells are measuring comparable conditions, corresponding to a water
table that is nearly recovered from seasonal pumping stresses. Over small times and distances,
these uncertainties can result in misleading results, but over longer times and distances they tend
to “average out,” resulting in a robust estimate of general trends.
Any program evaluating programs or managing resources at local or aquifer subunit level will
almost certainly need to obtain more, and possibly different, measurements than provided by the
107
annual monitoring program. Because we are pushing the data to its useful limit in trying to
evaluate water budgets over times of years at the township level, we list some of the major
possible perturbations of the data:
1. Interference by pumping – wells are not necessarily always shut down outside of the
irrigation season, and if a monitoring well or nearby wells have been recently pumped, an
artificially low water table will be measured – and the following year the water table will
show an apparent rise.
2. Incomplete recovery – even if all wells in an area have been off for the preceding four
months, the water table may not have fully recovered by early January. This has been
demonstrated in a variety of studies, and since the degree of recovery is likely to vary
from year to year, the relationship to a stable water table is a moving target.
3. Accuracy of land surface elevation – well elevations are estimated from a topographic
map and rarely can be considered to be accurate to better than + 5 feet. This has no effect
on differences measured in the same well, but if wells are added or replaced, there can be
a relative shift in local water table elevations. In addition, when elevations are used to
calculate hydrologic gradients to determine the direction and rate of groundwater flow,
errors in elevation can have a significant effect (see the calculations discussed below).
Water Extracted
Water extracted was determined from the KGS WIMAS database
(http://hercules.kgs.ku.edu/geohydro/wimas/index.cfm). Reported water use for the nine years
considered was extracted for each section with active water rights, and included in the update
database. Table D - 1 and Figure D - 4 show the township-level use value sums.
Table D - 1: Acre-feet per year of reported use for the four townships.
Twp/Year
9-32
9-33
9-34
10-33
ALL
1996
3993
7949
3720
3144
18806
1997
4164
9558
4307
3876
21905
1998
3786
7813
3568
3436
18602
1999
3326
6247
2706
2996
15276
108
2000
4738
11174
4724
4134
24771
2001
3498
8382
4063
3277
19220
2002
4504
10353
4985
4181
24022
2003
3884
9398
4540
3863
21685
2004
4275
9940
4548
4343
23106
Reported Water Use, AF/Township
12000
10000
8000
9-32
9-33
9-34
10-33
6000
4000
2000
0
1996
1997
1998
1999
2000
2001
2002
2003
2004
Figure D - 4: Reported water use for the four townships. Note that patterns of use are
very similar, and that there is a slight rising trend that counters the declining trend in
precipitation (Figure D - 5).
Although there is an uncertainty of about 20% in the relationship between the reported values
and the actual volume pumped (Hecox, 2003), the year-to year changes are probably quite
accurate on a relative basis. Overall, these are some of the best quantitative data that we have to
work with in the budgeting process. For comparison, if one inch of rain fell uniformly on a
standard-size township, the volume of water deposited would be 1920 AF. If this same amount
of water were transported to the water table with perfect efficiency and the aquifer had a specific
yield of 17%, the water table elevation would rise by ~0.59 ft. If there were not replacement for
any of the water shown above as pumped, the water table would be expected to fall about 2 ft./yr
under 9-33, and slightly more than 1 ft./year under the other townships.
Precipitation Data
Precipitation data for the three weather stations close to or in the area of interest are shown in
Table D - 2 and Figure D - 5 and Figure D - 6. All three stations show similar patterns, with no
systematic differences, so the average was applied to all four townships. Although most of the
precipitation returns to the atmosphere through evapotranspiration, the amount of precipitation
during and just before the growing season can influence water demand for irrigation during that
year, and precipitation is also a factor in determining the amounts of both natural and "enhanced"
recharge, which is discussed subsequently.
In general, most recharge originates with infiltration during the wettest years, and the increase or
decrease in water demand is typically seen most clearly in very dry or very wet years.
109
Table D - 2: Precipitation measurements in and near the Thomas County area of interest
(See Figure D - 1 for station locations).
YEAR
GROWING SEASON (MAR-OCT)
COLBY
1990
15.53
1991
15.9
1992
22.27
1993
26.24
1994
21.26
1995
21.37
1996
25.59
1997
18.4
1998
19.38
1999
18.64
2000
14.35
2001
15.42
2002
12.81
2003
13.85
2004
16.78
9604 avg
17.25
9604stdev
3.88
MINGO OAKLEY
16.43
17.91
17.69
20.63
26.87
22.36
16.12
18.81
18.94
16.38
17.16
22.33
18.16
16.84
17.83
18.34
20.17
18.97
16
15.9
16.88
20.25
9.49
14.39
12.89
10.7
16.44
16.84
16.11
17.17
3.16
3.39
ANNUAL
STD
DEV
1.20
1.27
1.16
2.44
2.57
2.50
4.25
0.84
0.79
0.81
0.93
2.48
2.50
1.61
0.22
AVG
16.62
16.80
21.45
25.16
18.73
18.90
21.69
17.80
18.52
19.26
15.42
17.52
12.23
12.48
16.69
16.84
3.08
STD
COLBY MINGO OAKLEY AVG DEV
18.12
18.68
20.88 19.23 1.46
18.81
21.08
19.95 1.61
26.24
23.91
25.08 1.65
30.79
29.71
25.12 28.54 3.01
24.42
18.72
21.51 21.55 2.85
22.22
20.05
17.72 20.00 2.25
26.09
18.17
23.42 22.56 4.03
20.19
21
19.02 20.07 1.00
22.41
21.68
22.79 22.29 0.56
19.32
21.11
19.87 20.10 0.92
16.37
18.24
18.16 17.59 1.06
18.61
20.11
22.64 20.45 2.04
13.7
9.72
15.05 12.82 2.77
14.52
13.61
11.91 13.35 1.32
20.07
19
19.83 19.63 0.56
19.03
18.07
19.19 18.76
3.87
3.96
3.79
3.54
Annual Precipitation (inches)
30
25
20
Colby
Mingo
15
Oakley
Average
10
5
0
1996
1997
1998
1999
2000
2001
2002
2003
2004
Figure D - 5: Total annual precipitation values for the years 1996-2004 for the three
weather stations in or near the area of interest. Note that the period preceding 1996 was
generally relatively wet (see Table D - 2).
110
Mar-Oct Precipitation (inches)
30
25
20
Colby
Mingo
15
Oakley
Average
10
5
0
1996
1997
1998
1999
2000
2001
2002
2003
2004
Figure D - 6: Rainfall during the months associated with the growing season, for the same
time periods as shown in Figure D - 3.
Groundwater Flow
Groundwater flow cannot be measured directly, but must be calculated from other data. Average
fluxes in and out of the pilot area were estimated for the nine-year period. We determined both
N-S and E-W water table gradients at each township boundary for each row or column of
sections, and summed the flows calculated from those results.
Input data included water levels and bedrock surface elevations from the KGS section-level
database and USGS hydraulic conductivity (which has been incorporated into the section-level
database).
Darcy’s Law was used to calculate groundwater flow with the following formula:
Q = KiA, where
Q = flow (ft3/day) and converted to (af/yr)
K = hydraulic conductivity (ft/day)
i = hydraulic gradient (unitless)
A = cross sectional area of saturated portion of aquifer (ft2).
Saturated thickness is the difference between the water table and bedrock elevations.
West to east and north to south cross sections were produced for all the rows and columns of
sections using the data mentioned above. The cross sections extend one township west, east, and
north of the pilot area, but only a short distance to the south because of lack of data. In addition
to the data listed above, predevelopment water table and land surface data from section-level
database were also obtained and plotted.
111
Figure D - 7 is an example of the cross sections. This section runs from west to east across the
area, in the sixth row of sections down from the north boundary, essentially across the center of
the area. It shows that saturated thickness is low in southwestern 9-34, that the gradient steepens
sharply at the boundary between 9-34 and 9-33, and that it then flattens out and remains that way
through 9-32. [See Appendix D3 for more data and example cross section figures.]
WE6
3400
LSE
WLE_PRE
WLE_AVG
BEDROCK_OGALLALA_UPDATED
3300
Elevation (ft)
3200
3100
3000
2900
9-34
9-33
9-32
09
S3
09 5W
S3 3 1
09 5W
S3 3 2
09 5W
S3 3 3
09 5W
S3 3 4
09 5W
S3 3 5
09 5W
S3 3 6
09 4W
S3 3 1
09 4W
S3 3 2
09 4W
S3 3 3
09 4W
S3 3 4
09 4W
S3 3 5
09 4W
S3 3 6
09 3W
S3 3 1
09 3W
S3 3 2
09 3W
S3 3 3
09 3W
S3 3 4
09 3W
S3 3 5
09 3W
S3 3 6
09 2W
S3 3 1
09 2W
S3 3 2
09 2W
S3 3 3
09 2W
S3 3 4
09 2W
S3 3 5
09 2W
S3 3 6
09 1W
S3 3 1
09 1W
S3 3 2
09 1W
S3 3 3
09 1W
S3 3 4
09 1W
S3 3 5
1W
36
2800
Figure D - 7: Example of west to east cross section showing elevations of land surface,
predevelopment water table, average 1996-2005 water table, and bedrock surface. See
Appendix D3 for the entire suite of cross section figures.
Hydraulic gradients across the west and east township boundaries were calculated primarily
based on the cross sections, but also considering the point water table measurements and
preliminary computer-generated contours (see example Figure D - 8; the red line is the
approximate location of the cross section shown in Figure D - 7). Estimates were made for each
row of sections across the west and east boundaries, and for the columns of sections across the
north and south boundaries. The summary results are tabulated for the townships separately and
as a group in Table D - 3.
It is important to realize that these calculations are made for a nearly recovered water table, and
reflect the overall equilibrium gradient appropriate for calculating long-distance transport. When
pumping occurs, local drawdown increases and re-orients short-range gradients. This will
112
accelerate and redirect flow at the section to township scale, but is likely to have little effect at
the township to county level.
Figure D - 8: Groundwater elevation contours. Groundwater flow is perpendicular to the
elevation lines, and the rate is typically faster where the contours are closer together. The
red line is the approximate location of the cross section shown in Figure D - 7. Note how
much influence one well can have on the shape of the contours (NW corner of 10-33).
Table D - 3: Summary of net groundwater fluxes. Positive numbers indicate a net inflow.
Negative numbers indicate a net outflow. See Appendix D3 for detailed data used to
produce the estimates.
9-34
9-33
9-32
10-33
ALL
Net groundwater flux (AF/yr)
-1307
3750
-331
2135
4247
113
Inflows and outflows (in AF/yr) for the townships are shown schematically in Figure D - 9
below. See Appendix D3 for detailed data used to produce the estimates.
145
511
4681
2256
9-34
379
2143
9-33
5101
973
9-32
2217
1951
281
314
256
10-33
3574
1750
567
Figure D - 9: Schematic illustration of the estimated groundwater flow relations among the
four Thomas County townships. The arrows with two flux values indicate the different
values obtained at either side of the respective township boundary, as a result of the
different hydraulic conductivities and saturated thicknesses.
Groundwater Flow Velocities
Specific discharges (Darcy velocities) and average linear velocities (“pore” velocities) were
calculated at the west and east boundaries of the area using the following equations.
Specific discharge = Ki
Average linear velocity = Ki/SY, where
SY is the USGS specific yield (incorporated into the KGS section-level database).
The average linear velocity (labeled “Pore” Velocity in Table D - 4) is an estimate of how fast a
particle of water will flow. Interim results in Table D - 4 indicate that groundwater flows at a
rate of about 1 ft/day, a number commonly used for the High Plains aquifer. Specific discharge
is a macroscopic concept to provide averaged descriptions of the pore behavior.
114
Table D - 4: Groundwater flow velocity estimates. The average linear velocity (labeled
“Pore” Velocity in the table) is an estimate of how fast a particle of water will flow.
WEST BOUNDARY (FLOW IN)
1996
Darcy Velocity
TRS
SY
(ft/d)
09S34W06 0.20
0.22
09S34W07 0.22
0.22
09S34W18 0.24
0.22
09S34W19 0.25
0.22
09S34W30 0.25
0.23
09S34W31 0.25
0.24
10S33W06 0.22
0.23
10S33W07 0.22
0.23
10S33W18 0.22
0.23
10S33W19 0.21
0.23
10S33W30 0.20
0.24
10S33W31 0.20
0.24
AVG_NORTH
AVG_10-33
AVG_ALL
EAST BOUNDARY (FLOW OUT)
1996
Darcy Velocity
TRS
SY
(ft/d)
09S32W01 0.20
0.14
09S32W12 0.20
0.12
09S32W13 0.17
0.12
09S32W24 0.15
0.11
09S32W25 0.14
0.10
09S32W36 0.13
0.08
10S33W01 0.16
0.09
10S33W12 0.17
0.09
10S33W13 0.18
0.09
10S33W24 0.19
0.09
10S33W25 0.18
0.09
10S33W36 0.16
0.09
AVG_NORTH
AVG_10-33
AVG_ALL
1996
"Pore" Velocity
(ft/d)
1.10
1.02
0.91
0.89
0.93
0.96
1.05
1.03
1.05
1.12
1.19
1.21
2005
Darcy Velocity
(ft/d)
0.23
0.23
0.23
0.23
0.23
0.23
0.23
0.23
0.23
0.23
0.24
0.24
0.97
1.11
1.04
1996
"Pore" Velocity
(ft/d)
0.71
0.60
0.71
0.75
0.71
0.63
0.57
0.51
0.49
0.46
0.48
0.53
0.68
0.51
0.60
2005
"Pore" Velocity
(ft/d)
1.14
1.03
0.97
0.94
0.93
0.92
1.05
1.06
1.07
1.12
1.19
1.21
0.99
1.12
1.05
2005
Darcy Velocity
(ft/d)
0.14
0.12
0.12
0.11
0.10
0.08
0.09
0.08
0.07
0.07
0.08
0.09
2005
"Pore" Velocity
(ft/d)
0.68
0.60
0.71
0.75
0.71
0.59
0.53
0.45
0.41
0.37
0.44
0.53
0.67
0.46
0.56
Based on the values tabulated above, the long term groundwater flow in the area takes
approximately 15-20 years per mile. While the effects of local pumping might speed this up
slightly, we consider it very unlikely that volume of ground water underneath a township could
115
be replaced in less than 50-60 years. This means that the first and greatest effects of either
conservation or depletion will be experienced in the immediate area.
Analysis and Discussion
Because water table changes and the volume of groundwater pumped are the only two
measurements we have that are directly relevant to the groundwater in storage, and because
pumping, in most areas, is expected to be the largest term in the groundwater budget, it is
important to determine how closely the two measures are related to each other and to
precipitation, the other variable for which we have direct measurements available.
Procedural note: In the analysis we have relied heavily on, and will frequently refer to, the
results of linear regression analysis (or, informally, ‘correlation’). Basically, this means
assuming that two variables (Y and X) are related by an equation for a straight line Y = mX +b,
where m is the slope and b is the intercept of the line (with the vertical or Y axis). We then test
how well this assumption is suited to our particular data, by determining not only m and b, but
also a statistic known as R (the correlation coefficient). R2 provides a measure of how well the
variability in one parameter is explained by the variability in the other. A perfect match
produces a value of 1.0, and a completely random relationship a value of zero.
In ‘real data’ determinations, and especially when environmental data of any kind are used, an R2
value > 0.9 indicates an extremely strong relationship, 0.7-0.8 are good correlations, and values
of 0.5-0.7 indicate that there probably is a relationship, but a noisy one. Still lower values may be
significant, but need to be interpreted with care and caution. Even good correlations do not
necessarily indicate direct relationships, however; if one variable controls another the correlation
will be good, but a good correlation doesn’t prove that there is a direct relationship. If two
variables are well correlated, they can be independent of each other, but both dependent on a
third variable (hence the saying “Correlation is not causation”). In our case, if pumping caused
water level change they should be well correlated, but that could also happen if precipitation had
a strong influence on both recharge and water demand. This is why we not only examine various
combinations of variables, but also consider the magnitudes and the values of m and b to see if
they make sense in terms of what we understand about hydrology.
Reported Use
Reported use was compared with water level change and precipitation, and precipitation and the
water table variables were also compared; results are shown in Appendix D2. Initially, the
section-center values obtained from TINs were used. Some degree of correlation was seen in
essentially all comparisons. The strongest correlations (highest R2 values) were between use and
water table change, with weaker correlation between use and precipitation. The seasonal
(March-October) precipitation value was typically better correlated than the annual.
The first group of use-change results were poorer for 10-33 than for the northern townships. We
examined the relationships between the well hydrographs and the water use data for the
townships (Figure D - 10abcd). There was substantial variation, both in the patterns and the
116
magnitudes of the water table changes, and in particular well 10-33-19cbd seemed questionable.
We removed it from the dataset and re-TINned the data (the new version of the dataset is
identified as “v2”); the results were moderately improved for 10-33, but still not impressive as
the cornerstone for the budget analysis.
In view of the variety of hydrographs seen among the wells that anchor the TINs used to
determine the water levels in the four townships, we decided to experiment with an alternative
approach. Rather than using the wells as geographic representatives, we tested various
combinations of the wells as multiple index wells by simply averaging the water level change
data and regressing that against the reported use. Optimum results were obtained with six wells
each for the northern townships, and three for 10-33. All of the new correlations showed
improvement over the use of the TIN values; some of them were substantial changes. The
graphs of the “multiple index” approach are shown in Figure D - 11, and can be compared with
the values using the TIN results, in Appendix D1. The change values in Figure D - 11 have been
converted to estimated AF by multiplying the feet of change by (640 x 36) acres/township, and
then by the USGS estimate of average specific yield for each township: 9-32, 17.3%; 9-33,
19.0%; 9-34, 20.9%; and 10-33, 19.1%. The plots in Appendix D1 are left in feet of change for
comparison purposes.
While we cannot be sure that the absolute elevations are any better, the multiple index well
average does a better job of relating the variations, and we have used those values in the rest of
the analysis. Where absolute elevations are required, as in the water flux determinations
described above, we have used the TINned v2 dataset.
117
Two of the six wells most
relevant to TWP 9-32
appear to each have two
annual measurements
that do not represent the
general water table
Figure D - 10a: Well hydrographs for the monitoring wells characterizing Township 9-32.
The reported water use plot is in the upper left. The water level changes are plotted with
increasing negative values toward the top of the plots; this is done so that variation is in the
same sense as the water use plots – higher use corresponds to greater declines. Points
marked in red are suspect with interannual changes of 9-12 ft. compared to the maximum
range of 1.5-3 ft. for other measurements.
118
Figure D - 10b: Hydrographs and water use plots for Township 9-33, arranged as in
Figure D - 10a. Wells indicated by dotted circles and arrows were not used in the
subsequent stage of analysis.
119
Figure D - 10c: Hydrographs and water use plots for Township 9-34. The two wells in 8-34
both showed sustained water level increases early in the period. Well 9-34-11 is the only
well in this area that is located close to the river channel, so the unusual positive change in
a year when most other wells were indicating some decrease needs to be considered in
terms of possible enhanced recharge under the channel (see Figure D - 11 and discussion
below).
120
Figure D - 10d: Water use and well hydrographs for Township 10-33. Dotted circles
indicate wells excluded from analysis for various reasons; for example, 10-34-12 shows
changes of 9-12 ft., compared to values no larger than 3-4 ft. in other wells. Dotted circles
indicate wells not used in the final average, and red indicates wells whose records showed
some sort of apparent anomaly (pattern or magnitude of change).
Figure D - 11 shows that the maximum annual decline calculated in acre-feet is somewhat larger
than the maximum reported use. The specific yield estimates may be inaccurate, but they are as
likely to be high as low, and are probably not off by more than 10-15% in any case (which is
generally less than the difference between use and change estimates). Since the data available
suggest that if unmetered use reports are systematically different from metered results, they are
likely to be higher, this suggests some distortion in the estimates of water table elevation.
Evidence for this has already been discussed above in reference to the occurrence of unusually
high annual change values. We suspect that some of these differences are due to the problem of
incomplete recovery. When pumping rates are high, the well will not have returned to as close as
usual to the equilibrium state, and the decline will be overestimated. If pumping rates are low or
normal in the following year, recovery will be more complete and the measurement will make up
the previous year's deficit and indicate lower than actual decline (or greater recovery) than is
actually the case. If any of the monitoring wells are affected by off-season pumping, the
difference can be even more striking.
121
9-32, CHNG vs RPT (AF)
index average
y = -7.166x + 24966
R2 = 0.7444
4000
9-34, CHNG vs RPT (AF)
index average
1000
2000
0
0
2500
-2000
y = -2.6775x + 8130.3
R2 = 0.828
2000
3000
3500
4000
4500
-10002000
5000
2500
3000
3500
4000
4500
5000
5500
- 2000
-4000
- 3000
-6000
- 4000
-8000
- 5000
- 6000
- 10000
- 7000
- 12000
9-33, CHNG vs RPT (AF)
index average
y = -2.6621x + 18808
R2 = 0.8331
4000
10-33, CHNG vs RPT (AF)
index average
y = -7.6892x + 22463
2
+22463
R = 0 588y = -7.R26892x
= 0. 588
2000
2000
0
0
-20005000
6000
7000
8000
9000
10000
11000
-20002000
12000
2500
3000
3500
4000
4500
-4000
-4000
-6000
-6000
-8000
-8000
- 10000
- 10000
- 12000
- 12000
- 14000
- 14000
- 16000
Figure D - 11: Annual change converted to estimated AF/Twp vs. reported use for each
township.
The plots in Figure D - 11 reveal a number of things about the apparent water budget when
examined in more detail. One obvious point is that the regression line crosses the zero-change
axis at about 7000 AF for 9-33 and 3000-3500 for the other townships. This would seem to
imply that this volume is the sustainable yield – with no decline when pumped at that rate.
However, the regression equation shows that about 2.7 AF is lost from storage for every AF
pumped from 9-33 and 9-34, and about 7.1 AF lost per AF pumped from 9-32 and 10-33. This is
clearly not a realistic possibility, especially since the addition of either precipitation or lateral
inflow to the system should make the ratio of storage loss to pumping <1 rather than >> 1.
Furthermore, the intercept values are the point at which the line would intercept the vertical
(change) or Y axis when the reported use axis is at zero (not shown on figure). This should
represent the long-term average inflow to the system (recharge plus other any other sources). If
we convert these volumes back to recharge (in inches at zero pumping), the values range from 4”
to 13”. However, the generally accepted long-term average value for recharge in the area is <1”.
Even with a generous allowance for enhanced recharge and lateral groundwater flow, the slope
and intercept values from the regression equation do not appear hydrologically reasonable.
A further point of interest is that 9-33, with about twice the pumping of the other townships,
shows about twice the apparent sustainability – presumably because more pumping produces
more signal distortion and less complete recovery in the water table response. However, there is
122
also another difference between 9-33 and the other townships, in that 9-33 has pumping
distributed rather uniformly, while the others all have significant areas with little or no pumping,
as shown in Figure D - 12. This may be introducing some systematic differences because of our
choice of treating all of the townships as township units, without considering the actual active
area for water use. This will be discussed in more detail later when we consider the combined
budget.
Average Use Density (2-mile radius), AF/section
35
34
33
32
31
S1
8
S3
S5
S7
9
S9
S11
S13
10
S15
S17
29
27
25
23
21
19
17
15
13
11
9
7
5
3
1
S19
500-550
450-500
400-450
350-400
300-350
250-300
200-250
150-200
100-150
50-100
0-50
Figure D - 12: Average 2-mile use density, AF/section. Areas less than ~ 50AF/section
experience little or no direct pumping stress, and may respond differently from the
pumped areas.
Recharge and Other Factors
Recharge is the term used to describe the addition of groundwater to a specific aquifer body from
some other compartment (we do not use it to describe lateral flow within the same aquifer unit).
The most common and usually the largest recharge component is usually rain or surface water
that percolates through the soil and eventually reaches the groundwater. Figure D - 13 illustrates
some of the factors affecting the amount of recharge.
123
Figure D - 13: Aspects of groundwater recharge. See text for discussion.
One component not shown in Figure D - 13 but illustrated in Figure D - 2 is addition of water
due to upward seepage from underlying geologic units. Hecox (2003) estimated from his model
studies of GMD 4 that about 10% of recent recharge might be due to this process over the entire
region. Because recharge is one of the smaller components of the water budget and 10% of it is
a very small component we are not specifically considering that source.
Regional recharge estimates are typically based on a very large scale water budget of the sort we
are attempting locally; the estimates we have available are those based on a USGS study that
estimated the average natural recharge rate in most of western Kansas as about 0.5” per year.
This would reflect the portion labeled natural recharge in Figure D - 13. Also natural, but locally
much larger, can be recharge beneath stream channels, especially if the water table is well below
the surface (a “losing stream”). Even if there is seldom flow in the channel, the natural
depression can act as a collector of water that flows over the surface or (more slowly) laterally
through the shallow soil layers.
Once the landscape is modified by cultivation, construction, etc., patterns and mechanisms of
recharge may change. Irrigation in particular is an important factor because not only can some of
the applied water return to the aquifer, but keeping the ground close to field capacity (the water
content at which drainage occurs) makes it more likely that precipitation falling on the moist soil
will generate recharge. Both of these provide enhancements to the natural recharge.
124
Most of the water (>90%) that arrives as precipitation returns to the atmosphere as
evapotranspiration. In western Kansas, runoff (surface flow) rarely moves much water for long
distances, except in extremely heavy storms or wet years. Wet periods in general are the major
source of groundwater recharge; during dry or normal years relatively little water may penetrate
below the rooting zone. However, when there is a thick unsaturated zone, water may take years
to decades to make the trip between surface and water table, so it is mostly the major events and
channel recharge that generate prompt responses of the water table.
The various components of recharge, and their spatial and temporal distributions, are difficult to
measure and challenging to estimate. Recharge is often the adjustment term used to close
budgets in which the other quantities are reasonably well known. In the estimates that follow we
use the USGS recharge estimates, averaged at the township level. As a long-term average these
would probably be low numbers since they do not take account of the human enhancements, but
it is not clear how well they represent the time period considered.
Water Budget
In order to estimate the combined water balance of the four townships, we used the reported
water use, the USGS recharge values, and the USGS specific yield estimates (for converting
water heights (thicknesses) or volumes to groundwater heights (thicknesses) or volumes). We
calculated the groundwater flux values at all four boundaries of each township, using the
equations and methods described above, the USGS hydraulic conductivity values, and water
table elevations derived from the v2 version of the updated section-level database. We derived
three versions of the water level change data: one based on the TIN process and the section level
database, another using the values obtained by averaging the records of the wells that provided
(as a group average) the best correlation with water use (see Figure D-10 above), and a third that
adjusted the second to reflect the fraction of the area of each township with an average annual 2mile use density > 50 AF/section (in this case the same adjustment was made to the recharge
values).
We used average values over the 9-year period to reduce the uncertainties and variability
associated with individual years. The results are presented in Table D - 5, where the first block of
numbers represents the water balance for each township and the total for each of the three
methods of calculating change. The second block breaks down the details of the overall net
groundwater flow numbers for each side of each township, and the third block presents the
component data used in the calculations.
We have not presented some of the efforts to examine individual years, but we found that
averaging the data to produce one budget was not only more efficient but much more credible in
terms of results than producing budgets from the individual year data and averaging those.
These findings, as well as the observation that a better balance can be achieved at the fourtownship level than for the individual townships, underscore the points made in the beginning:
the annual monitoring program is most reliably used over time scales of decades and spatial
scales >township; and more, better, and different data are needed for local understanding on an
annual basis.
125
Before discussing the implications of these findings for management and conservation, we need
to point out that the analysis used the change data that we identified above as "inflated" relative
to the actual withdrawals (slope, or ratio of CHNG/RPT >>1). This is one of the primary
arguments for the averaging process – although the individual annual change values plotted
(Figure D - 11) might be extremely divergent from a prediction based on use, the longer-term
average values presented in Table D - 5 are relatively well-behaved in terms of hydrologic
expectations – the change values are similar in magnitude but generally somewhat smaller than
the reported use values. The averaging process eliminates the extreme values without greatly
affecting the signal of the long-term trend.
126
Table D - 5: Summary of average budget term estimates for the period 1996-2004, calculated by: (1) using as water table changes
the average of the wells best correlated with use (Figure D - 10); (2) the TIN method applied to the database; and (3) as in (1),
but with the recharge and change values adjusted to reflect only the area of the township with used density >50 AF/section-year.
Twp
Use Report
(AF)
CHNG
(AF)
RCHG
(AF)
Flow
(Total, AF)
Net
Flow
(+ in, - out)
N
S
E
W
UD >50
fraction
AvgCHNG
(ft)
RECH (in)
SY (%)
HC (ft/day)
1. Multiple Index Well Average
9-34
9-33
9-32
10-33
ALL
9-34
-4481
-8986
-4243
-3823
-21533
-4481
-8986
-4243
-3823
-21533
-4481
-8986
-4243
-3823
-21533
4417
6639
2532
1534
15122
4185
5516
4623
2288
16612
3799
6639
2000
1212
13650
960
1018
1114
979
4071
960
1018
1114
979
8142
826
1018
880
773
3497
-1307
3750
-331
2135
4247
-1307
3750
-331
2135
4247
-1307
3750
-331
2135
4247
-59
2428
-704
954
2619
-643
1298
1163
1579
5304
-1164
2421
-1694
297
-139
145
973
-4681/
-5101
2256
1
-0.92
0.5
20.9
+3.3
101.1
+1.6
511
256/
281
-2143/
-1951
4681/
5101
1
2. TIN v2 Average CHNG
9-33
9-32
10-33
ALL
3. Area-Adjusted Index Well Values
9-34
9-33
9-32
10-33
ALL
-379
-256/
-281
At left
At left
At left
At left
At left
At left
At left
314
567
At left
At left
At left
At left
At left
At left
At left
-2217
-1750
At left
At left
At left
At left
At left
At left
At left
2143/
1951
3574
At left
At left
At left
At left
At left
At left
At left
1
1
1
1
1
1
.86
1
.79
.79
-1.51
-0.99
-0.35
0.53
+0.03
19.0
+1.8
98.6
+3.3
0.58
+0.02
17.3
+2.5
88.2
+7.5
0.51
+0.01
19.1
+1.7
89.4
+9.5
1
1
-1.05
-1.26
-0.96
-0.52
At left
At left
At left
At left
At left
At left
At left
At left
At left
At left
At left
At left
At left
At left
At left
At left
At left
At left
At left
At left
At left
At left
At left
At left
127
.86
If we have considered all the factors and the budget numbers are correct, the residuals in
the "Net" row should be zero. This appears not to be the case, although the residuals are
generally substantially smaller than the reported use term, and appreciably smaller than
the change term – the two parameters for which we have actual data. For the four
townships taken together as a system (ALL), the third option (area corrected multiple
index wells) appears best, with the uncorrected multiple index well approach better than
the TIN-based values. However, when uncertainty is considered, all of the budgets are
effectively "balanced." Table D - 6 shows the propagation of uncertainty through the
calculations for the TIN v2 result on the basis of two assumptions: the input data have an
uncertainty of either + 10% or + 20%. The first is unrealistically optimistic; the second
somewhat more realistic but still optimistic.
Table D - 6: Uncertainties in net water balance for the TIN v2case (see Table D - 5)
with two assumptions about the uncertainty of the input data.
Township
9-34
9-33
9-32
10-33
ALL
Net water in/out
(AF)
-643
1298
1163
1579
5304
Std. dev. 10% input
637
1131
649
528
2919
Std. dev. 20% input
1275
2262
1297
1056
5838
At + 20% uncertainty in the input data, the one standard deviation uncertainty for the
total budget and three of the four individual townships is >100% -- we simply cannot say
that these values are different from zero. At an unrealistically precise 10% input
uncertainty, the net uncertainty for two of the townships is close to 100 %, and for the
total, >50%.
Conservation and Management Implications
In spite of the high levels of uncertainty, some important conclusions can be reached
about the prospects for useful conservation programs in the area. One of the most
important points is that a significant fraction of the water being pumped comes from
recharge and lateral inflow rather than from groundwater storage (that is, decline). Based
on Table D - 5, we estimate this percentage as 30% for case 1, 23% for case 2 and 37%
for case 3. We use the middle value for an example.
If only 70% of the water pumped on average results directly in water table decline, a
cutback of pumping by 10% could yield up to a 14% decrease in the rate of water table
decline if the inflow and recharge were not significantly reduced. Changes in irrigation
would have essentially no effect on recharge from rainfall or runoff (channel recharge) or
in upward flow from the bedrock, and a 10% change would have only minor effects on
the irrigation-enhanced component of recharge. Similarly, we estimate that enhancement
of groundwater flow during the pumping season would amount to something like a 10%
128
change for a few months a year, so a fractional reduction in this would not be a major
factor.
To the extent that we can draw guidance from the water balance calculations, it appears
that the four townships in question are in a position to leverage conservation measures by
reducing groundwater declines by significantly more than the reduction in volume
pumped. The long-term groundwater pore velocities calculated above indicate that the
savings will have a primarily local effect for a period of at least several decades, so will
not be quickly lost to outflow. These same results apply to inflowing water too, however,
indicating that the rate at which pumped water is replenished is essentially as slow as the
rate at which conserved water is lost.
A significant factor in this high-leverage situation appears to be the large area of thin,
undeveloped aquifer to the west and south of the study region. It is probable that these
areas are functioning as recharge collectors, supporting the continued down-gradient flow
of groundwater at near-predevelopment rates, rather than being depleted as would be the
case if the area were developed to the same extent as the four townships. This supplies
water that is estimated to be a net positive (in greater than out) contribution of 2437
AF/yr for the four townships as a system.
Reducing Uncertainties: Data Needs
The somewhat positive findings outlined above are obviously approximate, and the data
used to arrive at them will not be adequate to monitor the effect of any conservation
programs on the time and space scales of interest. What is needed, and what are the
chances of reducing some of the uncertainties?
Water use reports are probably near + 20% in accuracy, and may be substantially better in
terms of precision. Improvement is possible and desirable, but would not address the
dominant uncertainties.
Recharge estimates can be improved, but the only component conveniently subject to
direct measurement on a routine basis is channel recharge – continuous or more frequent
measurements of wells near the base of the major stream channels, coordinated with local
rainfall and runoff or flow observations. Other aspects of recharge estimation can
probably be improved by models using data on soil types, ground cover, irrigated
acreage, etc.
Change in groundwater storage has two components:
Change in water table elevation is one of the biggest issues, and data on that can be very
substantially improved by
1. Measuring more wells more often (e.g., continuously or monthly) and with
particular attention to local conditions (e.g, current or recent pumping, etc.).
129
2. Selecting wells to monitor based on their suitability as areal indicators (the index
well approach) or as anchor points for the TIN process (the geographic sample
approach).
3. Surveying elevations of wells needed to link neighboring observations or
determine the absolute elevation (see 'later flow' below).
4. Improving 3-dimensional estimates of specific yield – which can be done by the
Practical Saturated Thickness estimation technique.
Lateral groundwater flow estimations can be improved by two actions:
1. Improved determination of the head gradient near and across the boundaries of
the area of interest. This will require both additional measurement points (which
can be simple and relatively inexpensive piezometers) and elevation surveys.
Some of the measurements need to be in the upgradient source area, since the
present water level measurement program is likely to systematically
underestimate water table elevations in unpumped areas (see Figure D - 14
below).
2. Estimates of the distribution of hydraulic conductivity (or actual flow) can be
improved to some extent by relatively simple methods (for example, using the
Practical Saturated Thickness determination process to assign approximate or
relative values to the strata identified.
The measures outlined above would in some cases improve our knowledge of the water
balance rather promptly, but the major effect would be felt after a few years of data
collection with a consistent system set up for the purpose at hand and designed and
measured at appropriate scales. The results will gradually permit us to calibrate on the
basis of field observations some of the parameters we cannot measure directly.
Once a larger inventory of better-qualified data is in, groundwater modeling at the local
scale can be useful in developing plans and testing hypotheses. At present, regional scale
models will not have the database or resolution to make accurate determinations in
transitional areas with limited data, and attempts to model at the local scale will suffer
from the same problems of data assessment and availability described in this report.
130
Figure D - 14: Monitoring water-level change in undeveloped areas – the upper
figure shows the natural gradient across regions of the aquifer with variable
saturated thickness. The areas of greater saturated thickness are developed; the
others are not (lower picture). Because water levels are preferentially monitored
where there are wells and pumping, post-development observations imply a uniform
reduction in water table elevation. However, the undeveloped area has its water
table supported by local recharge, and now becomes a source area, with water table
elevations higher than estimated from the monitoring wells.
Summary and Conclusions
The available information suggests that four-township region in southern Thomas County
has a significant amount of net groundwater inflow that enhances local recharge and
131
provides 'leverage' that increases the effectiveness of local groundwater conservation
measures in reducing declines.
This condition appears to be the result of their location adjacent to thin, undeveloped
areas of the aquifer that are funneling additional recharge into the region.
The movement of groundwater places limits on the rate at which the upgradient water is
supplied, but also ensures that any conservation benefits will remain beneath the four
townships for a substantial period of time.
The details of the water balance are uncertain because data on water levels and other
conditions in the region of interest are inadequate in both quantity and quality for
application at the time and space scales of concern.
A greatly improved data collection program to support management and assessment can
be put in place with a combination of one-time measurements and improvement with
expanded ongoing monitoring activities.
Reference
Hecox, G.R. 2003. GIS integration and error analysis for hydrogeologic evaluations.
Unpubl. Ph.D. Dissertation, Department of Geology, University of Kansas, Lawrence,
KS, 519p. Available as KGS Open-file Report 2003-53.
132
Appendix D-1 – v2 (Geographic Sample) Change vs. Use Regressions
9-32 v2
y = -0.0018x + 7.0764
R2 = 0.6019
1.5
1
CHNG ft
0.5
0
3000
-0.5
3500
4000
4500
5000
5500
-1
-1.5
-2
-2.5
-3
-3.5
RPT AF
y = -0.0005x + 2.9775
R2 = 0.7125
9-33 v2
0
5000
6000
7000
8000
9000
-0.5
CHNG, ft
-1
-1.5
-2
-2.5
-3
RPT AF
133
10000
11000
12000
y = -0.0012x + 3.997
R2 = 0.7563
9-34 v2
1
0.5
0
Chng, ft
2500
3000
3500
4000
4500
5000
5500
-0.5
-1
-1.5
-2
-2.5
RPT AF
y = -0.0019x + 6.9326
R2 = 0.3307
10-33 v2
3
2
Chang, ft
1
0
2500
3000
3500
4000
-1
-2
-3
-4
RPT AF
134
4500
5000
Appendix D 2
Water table elevation changes (TIN, data version 2) as a function of seasonal precipitation.
Seasonal precipitation consistently gave slightly higher R2 values than did annual.
However, the R2 values are generally poor, and the predominance of negative change
values means that precipitation, which by itself should have a zero or positive effect on
water level, is superimposed on the much stronger effect of decrease due to pumping.
WL CHANGE V SEASONAL PRECIP 9-32
WL CHNG V ANNUAL PRECIP 9-34
1.5
1
y = 0.1808x - 4.0067
R2 = 0.1744
1
y = 0.1739x - 4.3167
R2 = 0.407
0.5
0.5
0
0
0
5
10
15
20
0
25
- 0.5
5
10
15
20
25
-0.5
-1
-1
- 1.5
-2
-1.5
- 2.5
-2
-3
-2.5
- 3.5
WL CHNG V SEASON PRECIP 10-33
WL CHANGE V SEASONAL PRECIP 9-33
0
3
0
-0.5
-1
5
10
15
20
25
2
y = 0.1735x - 4.1798
2
R = 0.4674
y = 0.1694x - 3.4117
2
R = 0.1088
1
0
-1.5
0
-1
-2
-2
-2.5
-3
-3
-4
135
5
10
15
20
25
Reported use vs seasonal precipitation, by year and by township. R2 values are better
than with annual precipitation, but note how much the relationship would be affected if
a few of the highest or lowest values on either axis were dropped from the analysis.
RPT V SEASON PRECIP 9-32
RPT V SEASON PRECIP 9-34
5500
5500
5000
5000
4500
4500
4000
3500
4000
y = -80.39x + 5835.8
2
R = 0.1842
3000
y = -178.18x + 7244.6
2500
2
R = 0.6079
3500
3000
2000
10
15
20
25
10
RPT V SEASON PRECIP 9-33
15
20
25
RPT V SEASON PRECIP 10-33
12000
5000
11000
4500
10000
9000
4000
8000
3500
7000
3000y = -114.19x + 5801.5
R2 = 0.4214
2500
y = -329.17x + 14531
6000
R2 = 0.4445
5000
4000
2000
10
15
20
25
136
10
15
20
25
Appendix D-3
Contents
Detailed data used in groundwater flow estimates
West-East Cross Sections
North-South Cross Sections
1996 WEST BOUNDARY (FLOW IN)
TRS
09S34W06
09S34W07
09S34W18
09S34W19
09S34W30
09S34W31
10S33W06
10S33W07
10S33W18
10S33W19
10S33W30
10S33W31
ST96_V2
29.3
35.4
39.6
43.8
45.8
47.1
97.2
71.1
51.8
37.4
41.5
36.7
K (ft/d)
106
105
104
103
102
101
100
100
100
99
99
99
Q_IN_SUM
Q_IN_SUM
Q_IN_SUM
i
0.002083
0.00214
0.002102
0.002159
0.002273
0.002367
0.002311
0.002273
0.002311
0.002367
0.002405
0.002443
NORTH
10-33
ALL
Q
(ft3/d)
34172
41992
45681
51408
56088
59493
118633
85344
63231
46224
52195
46856
Q
(af/yr)
286
352
383
431
470
499
994
715
530
387
437
393
288833
412482
701315
2420
3456
5876
Q
(ft3/d)
57141
55184
56493
50243
40403
25867
37537
38308
39488
35254
35995
38431
Q
(af/yr)
479
462
473
421
339
217
315
321
331
295
302
322
285331
2391
1996 EAST BOUNDARY (FLOW OUT)
TRS
09S32W01
09S32W12
09S32W13
09S32W24
09S32W25
09S32W36
10S33W01
10S33W12
10S33W13
10S33W24
10S33W25
10S33W36
ST96_V2
76.3
87.6
89.0
84.3
77.3
59.9
77.5
83.0
84.4
75.8
79.2
85.6
K (ft/d)
96
94
92
89
83
72
85
81
78
75
71
67
Q_OUT_SUM
i
0.001477
0.001269
0.001307
0.001269
0.001193
0.001136
0.00108
0.00108
0.001136
0.001174
0.001212
0.001269
NORTH
137
Q_OUT_SUM
Q_OUT_SUM
IN MINUS
OUT
IN MINUS
OUT
IN MINUS
OUT
2005 WEST BOUNDARY (FLOW IN)
TRS
09S34W06
09S34W07
09S34W18
09S34W19
09S34W30
09S34W31
10S33W06
10S33W07
10S33W18
10S33W19
10S33W30
10S33W31
ST05_V2
21.5
27.1
30.7
34.3
37.5
40.4
83.6
68.0
50.0
36.5
41.4
37.4
10-33
ALL
225012
510344
1885
4276
3501
29
10-33
187469
1571
ALL
190971
1600
Q
(ft3/d)
26017
32384
37648
42438
45853
48977
102030
83632
61952
45143
52101
47780
Q(af/yr)
218
271
315
356
384
410
855
701
519
378
437
400
233316
392638
625954
1955
3290
5245
Q
(ft3/d)
48965
50431
50826
44069
35720
21257
31022
30701
30568
26621
31648
37311
Q(af/yr)
410
423
426
369
299
178
260
257
256
223
265
313
251268
187871
2105
1574
NORTH
K (ft/d)
106
105
104
103
102
101
100
100
100
99
99
99
Q_IN_SUM
Q_IN_SUM
Q_IN_SUM
i
0.002159
0.002159
0.002235
0.002273
0.002273
0.002273
0.002311
0.00233
0.002348
0.002367
0.002405
0.002443
NORTH
10-33
ALL
2005 EAST BOUNDARY (FLOW OUT)
TRS
09S32W01
09S32W12
09S32W13
09S32W24
09S32W25
09S32W36
10S33W01
10S33W12
10S33W13
10S33W24
10S33W25
10S33W36
ST05_V2
68.0
80.1
80.1
73.9
68.3
52.7
68.9
75.8
78.4
71.0
75.6
83.1
K (ft/d)
i
96
94
92
89
83
72
85
81
78
75
71
67
Q_OUT_SUM
Q_OUT_SUM
0.00142
0.001269
0.001307
0.001269
0.001193
0.001061
0.001004
0.000947
0.000947
0.000947
0.001117
0.001269
NORTH
10-33
138
Q_OUT_SUM
IN MINUS
OUT
IN MINUS
OUT
IN MINUS
OUT
•
ALL
439139
3680
NORTH
-17952
-150
10-33
204767
1716
ALL
186815
1565
West-East Cross sections labeled from north to south. For example Section
WE1 is across the northernmost row of sections, and WE12 is across the
southernmost.
West-East Section WE1
3400
3300
3200
3100
3000
2900
LSE
wte1996_v2
wte2005_v2
BEDROCK_OGALLALA_UPDATED
WLE_PRE
09
S3
09 5W
S3 0 6
09 5W
S3 0 5
09 5W
S3 0 4
09 5W
S3 0 3
09 5W
S3 0 2
09 5W
S3 0 1
09 4W
S3 0 6
09 4W
S3 0 5
09 4W
S3 0 4
09 4W
S3 0 3
09 4W
S3 0 2
09 4W
S3 0 1
09 3W
S3 0 6
09 3W
S3 0 5
09 3W
S3 0 4
09 3W
S3 0 3
09 3W
S3 0 2
09 3W
S3 0 1
09 2W
S3 0 6
09 2W
S3 0 5
09 2W
S3 0 4
09 2W
S3 0 3
09 2W
S3 0 2
09 2W
S3 0 1
09 1W
S3 0 6
09 1W
S3 0 5
09 1W
S3 0 4
09 1W
S3 0 3
09 1W
S3 0 2
1W
01
2800
139
S3
09 5W
S3 0 7
09 5W
S3 0 8
09 5W
S3 0 9
09 5W
S3 1 0
09 5W
S3 1 1
09 5W
S3 1 2
09 4W
S3 0 7
09 4W
S3 0 8
09 4W
S3 0 9
09 4W
S3 1 0
09 4W
S3 1 1
09 4W
S3 1 2
09 3W
S3 0 7
09 3W
S3 0 8
09 3W
S3 0 9
09 3W
S3 1 0
09 3W
S3 1 1
09 3W
S3 1 2
09 2W
S3 0 7
09 2W
S3 0 8
09 2W
S3 0 9
09 2W
S3 1 0
09 2W
S3 1 1
09 2W
S3 1 2
09 1W
S3 0 7
09 1W
S3 0 8
09 1W
S3 0 9
09 1W
S3 1 0
09 1W
S3 1 1
1W
12
09
West-East Section WE2
3400
3300
3200
3100
3000
2900
LSE
wte1996_v2
wte2005_v2
BEDROCK_OGALLALA_UPDATED
WLE_PRE
2800
140
S3
09 5W
S3 1 9
09 5W
S3 2 0
09 5W
S3 2 1
09 5W
S3 2 2
09 5W
S3 2 3
09 5W
S3 2 4
09 4W
S3 1 9
09 4W
S3 2 0
09 4W
S3 2 1
09 4W
S3 2 2
09 4W
S3 2 3
09 4W
S3 2 4
09 3W
S3 1 9
09 3W
S3 2 0
09 3W
S3 2 1
09 3W
S3 2 2
09 3W
S3 2 3
09 3W
S3 2 4
09 2W
S3 1 9
09 2W
S3 2 0
09 2W
S3 2 1
09 2W
S3 2 2
09 2W
S3 2 3
09 2W
S3 2 4
09 1W
S3 1 9
09 1W
S3 2 0
09 1W
S3 2 1
09 1W
S3 2 2
09 1W
S3 2 3
1W
24
09
09
S3
09 5W
S3 18
09 5W
S3 17
09 5W
S3 16
09 5W
S3 15
09 5W
S3 14
09 5W
S3 13
09 4W
S3 18
09 4W
S3 17
09 4W
S3 16
09 4W
S3 15
09 4W
S3 14
09 4W
S3 13
09 3W
S3 18
09 3W
S3 17
09 3W
S3 16
09 3W
S3 15
09 3W
S3 14
09 3W
S3 13
09 2W
S3 18
09 2W
S3 17
09 2W
S3 16
09 2W
S3 15
09 2W
S3 14
09 2W
S3 13
09 1W
S3 18
09 1W
S3 17
09 1W
S3 16
09 1W
S3 15
09 1W
S3 14
1W
13
West-East Section WE3
3400
3300
3200
3100
3000
2900
2900
LSE
wte1996_v2
wte2005_v2
BEDROCK_OGALLALA_UPDATED
WLE_PRE
2800
West-East Section WE4
3400
3300
3200
3100
3000
LSE
wte1996_v2
wte2005_v2
BEDROCK_OGALLALA_UPDATED
WLE_PRE
2800
141
09
S3
09 5W
S3 3 1
09 5W
S3 3 2
09 5W
S3 3 3
09 5W
S3 3 4
09 5W
S3 3 5
09 5W
S3 3 6
09 4W
S3 3 1
09 4W
S3 3 2
09 4W
S3 3 3
09 4W
S3 3 4
09 4W
S3 3 5
09 4W
S3 3 6
09 3W
S3 3 1
09 3W
S3 3 2
09 3W
S3 3 3
09 3W
S3 3 4
09 3W
S3 3 5
09 3W
S3 3 6
09 2W
S3 3 1
09 2W
S3 3 2
09 2W
S3 3 3
09 2W
S3 3 4
09 2W
S3 3 5
09 2W
S3 3 6
09 1W
S3 3 1
09 1W
S3 3 2
09 1W
S3 3 3
09 1W
S3 3 4
09 1W
S3 3 5
1W
36
Elevation (ft)
09
S3
09 5W
S3 3 0
09 5W
S3 2 9
09 5W
S3 2 8
09 5W
S3 2 7
09 5W
S3 2 6
09 5W
S3 2 5
09 4W
S3 3 0
09 4W
S3 2 9
09 4W
S3 2 8
09 4W
S3 2 7
09 4W
S3 2 6
09 4W
S3 2 5
09 3W
S3 3 0
09 3W
S3 2 9
09 3W
S3 2 8
09 3W
S3 2 7
09 3W
S3 2 6
09 3W
S3 2 5
09 2W
S3 3 0
09 2W
S3 2 9
09 2W
S3 2 8
09 2W
S3 2 7
09 2W
S3 2 6
09 2W
S3 2 5
09 1W
S3 3 0
09 1W
S3 2 9
09 1W
S3 2 8
09 1W
S3 2 7
09 1W
S3 2 6
1W
25
West-East Section WE5
3400
3300
3200
3100
3000
2900
LSE
WLE_PRE
wte1996_v2
wte2005_v2
BEDROCK_OGALLALA_UPDATED
2800
West-East Section WE6
3400
West Boundary
3300
LSE
wte1996_v2
wte2005_v2
BEDROCK_OGALLALA_UPDATED
WLE_PRE
3200
East Boundary
3100
3000
2900
2800
142
10
S3
10 4W
S3 0 6
10 4W
S3 0 5
10 4W
S3 0 4
10 4W
S3 0 3
10 4W
S3 0 2
10 4W
S3 0 1
10 3W
S3 0 6
10 3W
S3 0 5
10 3W
S3 0 4
10 3W
S3 0 3
10 3W
S3 0 2
10 3W
S3 0 1
10 2W
S3 0 6
10 2W
S3 0 5
10 2W
S3 0 4
10 2W
S3 0 3
10 2W
S3 0 2
2W
01
2900
2900
10
S3
10 4W
S3 0 7
10 4W
S3 0 8
10 4W
S3 0 9
10 4W
S3 1 0
10 4W
S3 1 1
10 4W
S3 1 2
10 3W
S3 0 7
10 3W
S3 0 8
10 3W
S3 0 9
10 3W
S3 1 0
10 3W
S3 1 1
10 3W
S3 1 2
10 2W
S3 0 7
10 2W
S3 0 8
10 2W
S3 0 9
10 2W
S3 1 0
10 2W
S3 1 1
2W
12
West-East Section WE7
3400
3300
3200
3100
3000
LSE
WLE_PRE
wte1996_v2
wte2005_v2
BEDROCK_OGALLALA_UPDATED
2800
West-East Section WE8
3400
3300
3200
3100
3000
LSE
wte1996_v2
wte2005_v2
BEDROCK_OGALLALA_UPDATED
WLE_PRE
2800
143
10
S3
10 4W
S3 1 9
10 4W
S3 2 0
10 4W
S3 2 1
10 4W
S3 2 2
10 4W
S3 2 3
10 4W
S3 2 4
10 3W
S3 1 9
10 3W
S3 2 0
10 3W
S3 2 1
10 3W
S3 2 2
10 3W
S3 2 3
10 3W
S3 2 4
10 2W
S3 1 9
10 2W
S3 2 0
10 2W
S3 2 1
10 2W
S3 2 2
10 2W
S3 2 3
2W
24
2900
2900
10
S3
10 4W
S3 3 1
10 4W
S3 3 2
10 4W
S3 3 3
10 4W
S3 3 4
10 4W
S3 3 5
10 4W
S3 3 6
10 3W
S3 3 1
10 3W
S3 3 2
10 3W
S3 3 3
10 3W
S3 3 4
10 3W
S3 3 5
10 3W
S3 3 6
10 2W
S3 3 1
10 2W
S3 3 2
10 2W
S3 3 3
10 2W
S3 3 4
10 2W
S3 3 5
2W
36
West-East Section WE10
3400
3300
3200
3100
3000
LSE
wte1996_v2
wte2005_v2
BEDROCK_OGALLALA_UPDATED
WLE_PRE
2800
West-East Section WE12
3400
3300
3200
3100
3000
LSE
wte1996_v2
wte2005_v2
BEDROCK_OGALLALA_UPDATED
WLE_PRE
2800
144
•
North-South Cross Sections. NS2 is across the west side of range 33; NS3 is
across the east side of range 33.
North-South Section NS2
3400
LSE
wte1996_v2
wte2005_v2
BEDROCK_OGALLALA_UPDATED
WLE_PRE
3300
3200
3100
3000
2900
10
S3
3W
31
10
S3
3W
30
10
S3
3W
19
10
S3
3W
18
10
S3
3W
07
10
S3
3W
06
31
09
S3
3W
09
S3
3W
30
09
S3
3W
19
09
S3
3W
18
09
S3
3W
07
09
S3
3W
06
2800
North-South Section NS3
3400
LSE
wte1996_v2
wte2005_v2
BEDROCK_OGALLALA_UPDATED
WLE_PRE
3300
3200
3100
3000
2900
145
3W
36
10
S3
3W
25
10
S3
3W
24
10
S3
3W
13
10
S3
3W
12
10
S3
10
S3
3W
01
3W
36
09
S3
3W
25
09
S3
3W
24
09
S3
3W
13
09
S3
3W
12
09
S3
09
S3
3W
01
2800
NS1 (west boundary) and NS4 (east boundary) – need to add WLE_PRE etc.
North-South Section NS1
3400
3300
3200
3100
3000
LSE
wte1996_v2
wte2005_v2
BEDROCK_OGALLALA_UPDATED
2900
4W
31
10
S3
4W
30
10
S3
4W
19
10
S3
4W
18
10
S3
4W
07
10
S3
10
S3
4W
06
4W
31
09
S3
4W
30
09
S3
4W
19
09
S3
4W
18
09
S3
4W
07
09
S3
09
S3
4W
06
2800
North-South Section NS4
3400
3300
LSE
wte1996_v2
wte2005_v2
BEDROCK_OGALLALA_UPDATED
3200
3100
3000
2900
2800
09S32W01 09S32W12 09S32W13 09S32W24 09S32W25 09S32W36 10S32W01 10S32W12 10S32W13 10S32W24 10S32W25 10S32W36
146
Appendix E: New Insights From Well Responses to Fluctuations In
Barometric Pressure
J.J. Butler, Jr.1,3, W. Jin1, G.A. Mohammed1,2, and E.C. Reboulet1
Accepted by Ground Water
Original Submission: November 2009
Revised Submission: August 2010
Accepted for Publication: September 2010
1) - Kansas Geological Survey, 1930 Constant Ave., Campus West, University of
Kansas, Lawrence, KS 66047, USA
2) - Dept. of Hydrology and Hydraulic Engineering, Vrije Universiteit Brussel, 1050
Brussels, Belgium. Now at Department of Geoscience, University of Calgary,
Calgary, Alberta, T2N 1N4, Canada
3) - Corresponding author – tel: 785-864-2116; email: [email protected]
147
ABSTRACT
Hydrologists have long recognized that changes in barometric pressure can produce
changes in water levels in wells. The barometric response function (BRF) has proven to
be an effective means to characterize this relationship; we show here how it can also be
utilized to glean valuable insights into semi-confined aquifer systems. The form of the
BRF indicates the degree of aquifer confinement, while a comparison of BRFs between
wells sheds light on hydrostratigraphic continuity. A new approach for estimating
hydraulic properties of aquitards from BRFs has been developed and verified. The BRF is
not an invariant characteristic of a well; in unconfined or semi-confined aquifers, it can
change with conditions in the vadose zone. Field data from a long-term research site
demonstrate the hydrostratigraphic insights that can be gained from monitoring water
levels and barometric pressure. Such insights should be of value for a wide range of
practical applications.
INTRODUCTION
For more than three centuries, scientists have known that changes in barometric
pressure can produce changes in water levels in wells (Pascal, 1973). Although the
phenomenon has long been recognized, the underlying mechanisms have only been
clarified much more recently (Jacob, 1940; Weeks, 1979; van der Kamp and Gale, 1983;
Rojstaczer, 1988; Spane, 2002). For a confined aquifer, a change in the barometric
pressure load on the land surface is transmitted downward, grain to grain, near
instantaneously to the interface between the confining unit and the aquifer. Part of the
load is then borne by the pore water and part is borne by the aquifer framework (Figure
148
22B in Ferris et al. [1962]). In contrast, the entire load is borne by the water column in a
well open to the atmosphere. The resulting pressure difference induces water flow
between the aquifer and the well, leading to the commonly observed inverse relationship
between barometric pressure and water level (Figure 1). The magnitude of the water-level
change primarily depends on how the load is shared between the pore water and the
aquifer framework, although the properties of the aquifer and overlying units, and the
characteristics of the well (e.g., well diameter and degree of well development) can also
play important roles. A different mechanism controls water-level responses in an
unconfined aquifer. In that case, access to the free water surface minimizes pore-pressure
changes produced by the grain-to-grain transmission of the surface load; the primary
control on responses is the downward propagation of air pressure through the pores of the
vadose zone (Figure 22A in Ferris et al. [1962]). For shallow water tables, this
propagation can occur so quickly that the pressure difference between the well and the
aquifer is negligible and, as a result, there is virtually no flow between the two. If the
propagation is delayed, due to the depth to water and/or conditions in the vadose zone,
the inverse relationship of Figure 1 is observed (Weeks, 1979; Hare and Morse, 1997;
Spane, 2002).
Hydrologists have traditionally characterized the relationship between barometric
pressure and water level using the ratio of the change in water level to the change in
barometric pressure head, which is termed the barometric efficiency (BE) and, by sign
convention, varies between zero and one (Jacob, 1940). In a confined aquifer, a BE value
near zero indicates that most of the load is borne by the pore water, while a value near
one indicates most is borne by the aquifer framework. Although BE has proven to be an
149
effective means of characterizing the short-term response of a well to a change in
barometric pressure, the barometric response function (BRF) is a more effective means
for characterizing the longer-term response (Rasmussen and Crawford, 1997; Spane,
2002). The BRF, which can be determined through a regression convolution procedure
(Furbish, 1991; Rasmussen and Crawford, 1997; Toll and Rasmussen, 2007),
characterizes the water-level response over time to a step change in barometric pressure,
essentially BE as a function of time since the imposed load. The BRF has been
successfully used to remove the effect of barometric-pressure changes on water levels
(Toll and Rasmussen, 2007), a critical step, for example, in the interpretation of pumping
tests when drawdown is small (Batu, 1998). Rasmussen and Crawford (1997) and Spane
(2002) discuss the impact of well conditions and site hydrogeology on BRFs and propose
characteristic BRF forms for certain hydrogeologic settings (confined and deep
unconfined aquifers). Spane (2002) reviews a number of time- (e.g., Toll and Rasmussen,
2007) and frequency- (e.g., Quilty and Roeloffs, 1991) domain approaches that have been
proposed for removing the effect of barometric-pressure changes from water-level data
series and concludes that the BRF (a time-domain method) is particularly effective for
this purpose.
Water-level responses to fluctuations in barometric pressure have also been used to
estimate subsurface properties. Specific storage can be determined from BE if estimates
of aquifer porosity and pore-water compressibility are available (Jacob, 1940; Batu,
1998). Time- and frequency-domain methods, often in a type-curve format, have been
developed for determining hydraulic properties from water-level responses to barometricpressure changes (e.g., Weeks, 1978; Rojstaczer, 1988; Rojstaczer and Riley, 1990;
150
Evans et al., 1991; Furbish, 1991). These methods, however, have yet to be widely
adopted.
The purpose of this paper is to extend the earlier work to show how the BRF can be
utilized to glean further hydrogeologic insights. Our primary emphasis is on gaining
insights into the low permeability unit (henceforth, aquitard) that overlies a semiconfined aquifer. Given the utility of the BRF for removing the impact of barometricpressure changes from water-level data, we also explore its value for estimating
subsurface properties. We propose a new approach for estimating aquitard hydraulic
conductivity (K) by fitting theoretical responses to field-determined BRFs. We
demonstrate these concepts using field data from a long-term research site and discuss
how the BRF can also be used to gain insights into conditions in unconfined aquifers and,
potentially, the vadose zone. Although BE is considered an invariant parameter of a well,
we show that a BRF can change as a function of conditions in the vadose zone.
Field Site Overview
The field component of this study took place at the Larned Research Site (LRS;
38.2◦ N latitude, 99.0◦ W longitude) of the Kansas Geological Survey (Figure 2a). The
primary focus here is on three LRS wells (LWC2, LEA5, and LEC2 in Figure 2a; all
0.102-m inner diameter) screened in the semi-confined High Plains aquifer (interval A in
Figure 2b), with a secondary focus on adjacent wells screened at the bottom of the
unconfined Arkansas River alluvial aquifer (interval B in Figure 2b). Each well has an
integrated pressure-transducer and datalogger unit submerged in the water column
(miniTroll, In-Situ, Inc.); pressure readings are taken every 15 minutes. Gauge (relative
151
to atmospheric pressure) and absolute pressure transducers were used in this work; Price
(2009) describes how both types of transducers can be utilized for assessing water-level
responses to barometric-pressure changes. Atmospheric pressure is recorded with on-site
barometers at the top of wells (baroTroll, In-Situ, Inc.) and at a weather station (Onset
Computing Corp.); readings are also taken every 15 minutes. Groundwater in the vicinity
of the LRS is primarily used for irrigation, so the vast majority of pumping is during the
growing season (mid-March to mid-October). In most years, water levels recover from
seasonal pumping by mid-December (Figure 3). The Arkansas River, which was flowing
at the time of the photograph in Figure 2a, is intermittent at the LRS and had little to no
flow for the period of the analyses discussed here. Note that all of the LRS wells in the
semi-confined High Plains aquifer display a pronounced water-level response to changes
in barometric pressure (e.g., Figure 1). Small responses (a few mm or less) to
precipitation loading (e.g., Rasmussen and Mote, 2007) have been observed, but had no
influence on the analyses. Water-level responses to stream-stage loading (e.g., Boutt,
2010) have also been observed, but were negligible during the period of the analyses.
Methodology
BRFs were determined from the LRS water-level and barometric-pressure data
using the regression convolution approach of Furbish (1991) and Rasmussen and
Crawford (1997). This approach, which has been implemented in a spreadsheet format
(e.g., Halford, 2006; Toll and Rasmussen 2007), assumes that temporal changes in a
detrended (removal of linear trend in this work) time series of water levels (equally
spaced in time) can be represented as
152
m
n
i =0
i =0
∆ W ( t ) = ∑ α i ∆ B (t − i ∆ t ) + ∑ β i ∆ E ( t − i ∆ t )
(1)
where ∆W(t) is the change in detrended water-level elevation [L] between time t and the
previous time when a measurement was taken (t-∆t); ∆B(t-i∆t) and ∆E(t-i∆t) are the
changes in the detrended barometric-pressure head [L] and earth-tide gravity potential
[LT-2], respectively, between time t-i∆t and the previous time when a measurement was
taken [t-(i+1)∆t]; αi and βi are the unit (impulse) barometric-pressure and earth-tide
response functions at lag i, respectively; m is the maximum time lag for the barometric
pressure response; n is the maximum time lag for the earth tide response; and ∆t is the
time between adjacent measurements. The underlying assumption of this implementation
of the BRF approach is that ∆W(t) is only a function of changes in barometric pressure
and the earth-tide gravity potential, i.e. the impact of other mechanisms on water levels is
negligible or can be removed by detrending the water-level data. This assumption appears
quite reasonable for systems such as the High Plains aquifer in western Kansas where
recharge is very small and pumping is seasonal in nature. The earth-tide gravity potentials
for the LRS are obtained with TSOFT, which generates synthetic earth tide records for a
given location (Van Camp and Vauterin, 2005). Earth tides do have a small effect on
water levels at the LRS, so they are incorporated in the analysis following the approach
outlined in Toll and Rasmussen (2007). The focus of this paper, however, is on the much
larger fluctuations induced by changes in barometric pressure and the insights that can be
derived from them.
153
Ordinary least-squares linear regression is used to estimate αi and βi, and the
barometric response function for lag j, Aj, is obtained by summing the αi terms up to that
lag:
j
A j = ∑α i
(2)
i =0
with the standard error given as
σ̂ Aj =
j
j
∑∑ C
i =0 k = 0
(3)
i ,k
where C is the variance-covariance matrix for the αi estimates (e.g., Abraham and
Ledolter, 1983). A small BE and finite transducer resolution can result in occasions when
∆W(t) is incorrectly truncated to zero. In order to reduce such truncation errors, the above
approach can be extended to incorporate water-level and barometric-pressure changes
over multiples of ∆t.
We have developed theoretical BRFs for an aquifer system similar to that of Figure
2b using a semi-analytical solution for a 1-D vertical representation of a two-layer
(aquitard and aquifer – see vertical bar on Figure 2b) configuration. The governing
equations, which are based on the development of van der Kamp and Gale (1983), are
∂h1
∂ 2 h1
− γ 1 h0 δ(t ) = D1
∂t
∂z 2
(4a)
∂h 2
∂ 2 h2
− γ 2 h0 δ(t ) = D2
∂t
∂z 2
(4b)
where h0 is the change in barometric pressure head at the land surface [L], δ(t) is the delta
function [T-1], and hi, Di, and γi are head deviation from static [L], hydraulic diffusivity
[L2T-1], and loading efficiency (1-BE) [-] for the aquitard (1) and aquifer (2),
154
respectively, and z is depth (0 at aquifer-aquitard interface and increases downward). The
hydraulic diffusivity is the ratio of hydraulic conductivity (Ki, [LT-1]) over specific
storage (Ssi, [L-1]). The loading efficiency term (γih0δ(t)) represents the pressurization of
the pore water via the near-instantaneous grain-to-grain transmission downward of the
surface load. Groundwater flow is primarily driven by the boundary condition at the top
of the aquitard, which is a function of the pressure propagation through the pores of the
overlying vadose zone and unconfined aquifer.
The initial condition for the system is static heads in the aquifer and aquitard (i.e., hi
is zero); the boundary conditions are a constant head at the top of the aquitard (produced
by the propagation of a step change in barometric pressure head to the bottom of the
overlying unconfined aquifer), zero flow at the bottom of the aquifer, and continuity of
head and flow at the aquitard-aquifer interface.
A solution for the governing equations, (4a)-(4b), and auxiliary conditions is
obtained using standard integral-transform techniques. The system of equations is
transformed into Laplace space and solved to yield the transform-space solution:
hi (z , p ) =
h0
f i (z , p )
p
where hi is the Laplace transform of hi, p is the Laplace-transform variable,
155
(5a)
f1 (z, p ) =
⎛ p ⎞
⎛ p ⎞
⎛ p ⎞
⎛ hUB
⎞
D2
⎜⎜
sech⎜⎜
l ⎟⎟ − (γ 1 − γ 2 ) tanh ⎜⎜
l ⎟⎟ tanh ⎜⎜
a ⎟⎟
− γ 1 ⎟⎟ K r
h
D
D
D
D
⎛ p ⎞
1
1 ⎠
1 ⎠
2 ⎠
⎝ 0
⎠
⎝
⎝
⎝
cosh ⎜⎜
z ⎟⎟
D
⎛ p ⎞
⎛ p ⎞
D2
1
⎠
⎝
Kr
l ⎟⎟ tanh ⎜⎜
a ⎟⎟
+ tanh ⎜⎜
D1
D
D
1 ⎠
2 ⎠
⎝
⎝
⎛ p ⎞
⎛ hUB
⎞
⎜⎜
l ⎟ + (γ 1 − γ 2 )
− γ 1 ⎟⎟sech⎜⎜
D1 ⎟⎠
⎛ p ⎞
⎛ p ⎞
⎝ h0
⎠
⎝
tanh ⎜⎜
a ⎟⎟ sinh ⎜⎜
z ⎟⎟
−
D
D
⎛ p ⎞
⎛ p ⎞
D2
2 ⎠
1 ⎠
⎝
⎝
Kr
l ⎟⎟ tanh ⎜⎜
a ⎟⎟
+ tanh ⎜⎜
D1
⎝ D2 ⎠
⎝ D1 ⎠
+ γ1
(5b)
for the aquitard (−l ≤ z ≤ 0),
⎤
⎡ p
⎛ p ⎞
⎛ hUB
⎞
⎜⎜
(z − a )⎥
l ⎟⎟ + (γ 1 − γ 2 ) cosh ⎢
− γ 1 ⎟⎟sech⎜⎜
⎝ h0
⎠
⎦ +γ
⎣ D2
⎝ D1 ⎠
f 2 (z, p ) =
2
⎛ p ⎞
⎛ p ⎞
⎛ p ⎞
1 D1
1+
tanh ⎜⎜
l ⎟⎟ tanh ⎜⎜
a ⎟⎟ cosh ⎜⎜
a ⎟⎟
K r D2
⎝ D2 ⎠
⎝ D2 ⎠
⎝ D1 ⎠
(5c)
for the aquifer (0 ≤ z ≤ a), Kr is K1/K2, hUB is the constant head at the top of the aquitard,
a is aquifer thickness, and l is aquitard thickness. The derivation of equation (5a) is given
in the Appendix.
The real-space form of equation (5a) is generated using the inversion algorithm of
Stehfest (1970). The expression for the head in the semi-confined aquifer is
V n ⎛ ln 2 ⎞
f 2 ⎜ z,
n⎟
(6)
t ⎠
⎝
n =1 n
where Vn is the coefficient for the Laplace inversion and N is the number of terms in the
N
h2 (z , t ) ≈ h0 ∑
Stehfest summation (14 for this work). The barometric response function for a well in the
semi-confined aquifer is
156
N
V
h2 (z, t )
⎛ ln 2 ⎞
≈ 1 − ∑ n f 2 ⎜ z,
A(z , t ) = 1 −
n⎟
(7)
h0
t ⎠
⎝
n =1 n
Equation (7) assumes a constant head (hUB) at the top of the aquitard. Temporal variations
in that head can be readily incorporated using superposition (convolution) procedures
(e.g., Olsthoorn, 2008) as shown in the Appendix. Although wellbore storage is ignored
in this development because of the rapid (relative to the typical ∆t used in practice)
response of wells in aquifers of moderate to high K, the solution can be extended to
incorporate wellbore storage following Furbish (1991) and Spane (2002). Similarly, the
solution can be extended to incorporate the vadose zone following Weeks (1979) and
others.
Application
The regression convolution approach was applied to data from three LRS wells
(LWC2, LEA5, and LEC2 – Figure 2a) and the site reference barometer (adjacent to
LEC2). Winter 2003-04 (henceforth, winter 2004) data were used because there was
virtually no pumping then and well responses appeared to be representative of typical
conditions observed in LRS High Plains aquifer wells (Figure 3). The winter 2004 BRFs
(Figure 4) have three important characteristics. First, the agreement between the BRFs
from the different wells is quite striking, despite the wells being separated by over 680 m,
indicating that the character of the aquifer-aquitard system is not changing substantially
between the wells. Second, the short-term (one-hour) response is typical of what would
be expected in a confined aquifer in which most of the load is borne by the pore water
(BE≈0.08), consistent with the near-surface, unconsolidated nature of the aquifer (e.g.,
Rasmussen and Mote, 2007). Third, the longer-term (one-day) response is typical of what
157
would be expected in a semi-confined (leaky) aquifer where water movement through the
aquitard equilibrates heads, consistent with the results of a four-day pumping test at the
LRS in which drawdown stabilized as a result of leakage (Butler et al., 2004). The BRFs
of wells screened at the bottom of the unconfined aquifer (interval B of Figure 2b) were
zero for this time period (winter 2004 curve of Figure 5), indicating that barometricpressure changes propagated rapidly through the pores of the vadose zone and the
unconfined aquifer. The rapid propagation across the unconfined aquifer (BRFs from
LRS wells screened at the water table and those screened at the bottom of the unconfined
aquifer always coincide) indicates that the apparent clay layers in the unconfined aquifer
shown in the EC log of Figure 2b are not laterally extensive enough to affect the
hydraulic connection between the top and bottom of that aquifer for the temporal
resolution (∆t = 15 min) of this analysis.
The BRFs presented in Figure 4 suggest the possibility of acquiring information
about the aquitard from these functions. Theoretical response functions were computed
using equation (7) and fit to the field-determined BRFs to estimate the properties of the
aquifer-aquitard system. Conditions at the top of the aquitard, which are required for the
response function calculation, were obtained from wells screened at the bottom of the
unconfined aquifer (interval B of Figure 2b). For winter 2004, the BRFs for those wells
were essentially zero for all lags beyond the zero lag (winter 2004 curve of Figure 5).
Using that condition at the aquitard top, an aquifer loading efficiency (1-BE) of 0.92, and
an estimate of aquifer diffusivity (2.9 × 106 m2d-1) based on previous estimates of aquifer
K (88 md-1) and Ss (3.0 × 10-5 m-1) obtained from the four-day LRS pumping test (Butler
et al., 2004), we fit a theoretical response function to the winter 2004 BRF for well LEA5
158
(Figure 6). The fit, which was based on the first half-day of the BRF because additional
mechanisms appear to be affecting the BRF at larger times, yielded estimates of the
aquitard diffusivity (D1 =1.7 × 102 m2d-1 ), the aquitard loading efficiency (γ1 = 0.97), and
the ratio between the aquifer and aquitard hydraulic conductivity (Kr = 1.8 × 10-5). Using
the K2 estimate from the LRS pumping test, an aquitard K of 1.6 × 10-3 md-1, which is
within 25% of the pumping-test value (2.1 × 10-3 md-1), is calculated from the Kr
estimate, demonstrating the similarity of the Kr ratios obtained with the different
approaches. Note that an estimate of aquifer diffusivity is required for the parameter
estimation procedure because of the high degree of correlation between Kr and D2. In the
absence of the pumping-test information that was available at the LRS, the aquifer
diffusivity could be estimated using the aquifer K from a slug test and the aquifer Ss
determined from the BE. In this example, the target for comparison was the aquitard K
from the LRS pumping test, so the aquifer K and Ss values from that same pumping test
were used for the diffusivity estimate. Isotropy in aquifer hydraulic conductivity was
assumed for the calculation of K1. This assumption should be reasonable in
unconsolidated aquifers with a hydrostratigraphic framework similar to that at the LRS
(Figure 2b).
A check on the parameters calculated from the BRF fit was performed by using the
same parameters to generate a theoretical response function to compare with the fielddetermined BRF for winter 2008; this time was chosen because it followed an extended
period of recharge (Figure 3). Although intuitively one might expect the BRF to be a
characteristic function of a well, our results show otherwise. The BRF for well LEA5 in
winter 2008 (Figure 6) is distinctly different from the 2004 BRF because of differing
159
conditions at the top of the aquitard (bottom of the unconfined aquifer). In winter 2008,
the wells screened at the bottom of the unconfined aquifer display a barometric response
(e.g., winter 2008 BRF in Figure 5), indicating that the propagation of air pressure
through the vadose zone was affected by a change of conditions in that zone (e.g.,
perched water table or layer of frozen soil). This difference in barometric responses at the
bottom of the unconfined aquifer between 2004 and 2008 is analogous to the difference
reported by Hare and Morse (1997) between a well below a low permeability landfill cap
and one adjacent to the cap. Moreover, the 2008 response in the unconfined aquifer
cannot be matched with the one-dimensional, uniform vadose zone solution of Weeks
(1979), indicating that the response must be a product of complicated flow paths or other
phenomena. Using the parameters determined from the 2004 fit and the upper boundary
condition based on the 2008 data (i.e. the winter 2008 BRF of Figure 5), we generated the
2008 theoretical response function for well LEA5 shown in Figure 6. The agreement with
the 2008 field-determined BRF is quite good, although no curve fitting was involved, and
can be considered a strong verification of the parameters calculated from the 2004
analysis.
Discussions and Conclusions
This work demonstrates the utility of the barometric response function (BRF) for
gaining insights into site hydrostratigraphy. In semi-confined aquifers, the form of the
BRF indicates the degree of aquifer confinement and can be exploited to estimate
aquitard properties using the approach developed here. A comparison of BRFs between
wells can shed light on aquitard continuity. However, the generality of the conclusions
160
that can be drawn from this comparison depends on the BRF averaging (support) volume,
which is the subject of ongoing work. In unconfined aquifers, the similarity of BRFs from
wells screened at the water table and those at the base of the aquifer indicate that low-K
layers within the aquifer are not laterally extensive enough to affect the hydraulic
connection across the aquifer for the temporal resolution of this analysis. Differences
between such BRFs could potentially be exploited to estimate the vertical K of an
unconfined aquifer in a manner analogous to the frequency-domain approach of
Rojstaczer and Riley (1990).
This work appears to be the first to show that BRFs are not necessarily an invariant
characteristic of a well. The form of a BRF can depend on the nature of the pressure
propagation through the vadose zone, even for wells in a semi-confined aquifer (Figure
6). This dependence on the vadose zone presents the opportunity to glean insights into
changes in vadose-zone conditions, a subject of ongoing work. Spane (2002) and others
have speculated that the barometric response of wells in unconfined aquifers could vary
as a function of vadose-zone conditions. This work confirms that speculation and
demonstrates that a similar dependence is found in semi-confined aquifers (Figures 5-6).
The barometric response function is a promising tool for gaining important insights
through monitoring of water levels and barometric pressure. In this study, we
demonstrated that BRFs can provide reasonable estimates of aquitard properties as well
as valuable information about other aspects of site hydrostratigraphy. They thus can often
be a cost-effective alternative/supplement to a conventional pumping test. Similarly,
BRFs should be a useful tool for initial screening of potential shallow target zones for
CO2 sequestration and, more generally, for monitoring changes in formation and fluid
161
properties (e.g., porosity and fluid compressibility) in the course of sequestration activities.
Although we demonstrated the approach in a system in which the strong seasonality of
pumping facilitated data processing, it is also applicable in aquifers that are pumped more
continuously, although more involved processing is required to remove the impact of
other mechanisms. Finally, we must emphasize that the approach for estimation of
aquitard properties described here is best implemented with a well in the overlying
aquifer. In the absence of such a well, considerable error may be introduced into the
parameter estimates through uncertainty about head conditions at the top of the aquitard.
This uncertainty can be particularly large at sites with thick vadose zones.
ACKNOWLEDGMENTS
This paper greatly benefited from reviews provided by Todd Rasmussen, Garth
van der Kamp, Frank Spane, Geoff Bohling, Peter Dietrich, and an anonymous reviewer.
GAM was a 2008 participant in the Applied Geohydrology Summer Research
Assistantship program of the Kansas Geological Survey. This program is open to
students at any university with an interest in learning more about recent developments in
hydrogeological field methods.
162
REFERENCES
Abraham B., and J. Ledolter, 1983, Statistical Methods for Forecasting, Wiley.
Batu, V. 1998. Aquifer Hydraulics. John Wiley and Sons, New York.
Boutt, D.F. 2010. Poroelastic loading of an aquifer due to upstream dam releases. Ground
Water 48, no. 4:580-592.
Butler, J.J., Jr., D.O. Whittemore, X. Zhan, and J.M. Healey. 2004. Analysis of two
pumping tests at the O’Rourke Bridge site on the Arkansas River in Pawnee County,
Kansas. Kansas Geological Survey Open-File Rept. 2004-32.
Evans, K., J. Beavan, D. Simpson, and S. Mousa. 1991. Estimating aquifer parameters
from analysis of forced fluctuations in well level: An example from the Nubian
formation near Aswan, Egypt 3. Diffusivity estimates for saturated and unsaturated
zones. J. Geophys. Res. 96, no. B7:12,161-12,191.
Ferris, J.G., D.B. Knowles, R.H. Brown, and R.W. Stallman. 1962. Theory of aquifer
tests. U.S. Geological Survey Water-Supply Paper 1536-E
(http://pubs.usgs.gov/wsp/wsp1536-E/pdf/wsp_1536-E_a.pdf).
Furbish D.J. 1991. The response of water level in a well to a time series of atmospheric
loading under confined conditions. Water Resour. Res. 27, no. 4:557-568.
Halford K.J. 2006. Documentation of a spreadsheet for time-series analysis and
drawdown estimation. U.S. Geological Survey Scientific Investigations Report 20065024.
Hare, P.W. and R.E. Morse. 1997. Water-level fluctuations due to barometric pressure
changes in an isolated portion of an unconfined aquifer. Ground Water 35, no. 4:667671.
163
Jacob, C.E. 1940. On the flow of water in an elastic artesian aquifer. Trans. AGU 21:574586.
Olsthoorn, T.N. 2008. Do a bit more with convolution. Ground Water 46, no. 1:13-22.
Pascal, B. 1973. The Physical Treatises of Pascal. Octagon Books, New York.
Price, M. 2009. Barometric water-level fluctuations and their measurement using vented
and non-vented pressure transducers. Quarterly Jour. of Engineering Geology and
Hydrogeology 42, no. 2: 245-250.
Quilty, E.G. and E.A. Roeloffs. 1991. Removal of barometric pressure response from
water level data. J. Geophys. Res. 96, no. B6: 10,209-10,218.
Rasmussen, T.C. and L.A. Crawford. 1997. Identifying and removing barometric
pressure effects in confined and unconfined aquifers. Ground Water 35, no. 3: 502–
511.
Rasmussen, T.C. and T.L. Mote. 2007. Monitoring surface and subsurface water storage
using confined aquifer water levels at the Savannah River site, USA. Vadose Zone
Jour. 6:327-335.
Rojstaczer, S. 1988. Determination of fluid flow properties from the response of water
levels in wells to atmospheric loading. Water Resour. Res. 24, no. 11: 1927-1938.
Rojstaczer, S. and F.S. Riley. 1990. Response of the water level in a well to earth tides
and atmospheric loading under unconfined conditions. Water Resour. Res. 26, no. 8:
1803-1817.
Spane, F.A. 2002. Considering barometric pressure in ground-water flow investigations.
Water Resour. Res. 38, no. 6: 10.1029/2001WR000701.
164
Stehfest, H. 1970. Numerical inversion of Laplace transforms. Commun. ACM. 13, no. 1:
47–49.
Toll, N.J. and T.C. Rasmussen. 2007. Removal of barometric pressure effects and earth
tides from observed water levels. Ground Water 45, no. 1: 101–105.
Van Camp, M., and P. Vauterin. 2005. Tsoft: graphical and interactive software for the
analysis of time series and Earth tides. Computers & Geosciences 31, no. 5: 631-640.
van der Kamp, G., and J.E. Gale. 1983. Theory of earth tide and barometric effects in
porous formations with compressible grains. Water Resour. Res. 19, no. 2: 538-544.
Weeks, E.P. 1979. Barometric fluctuations in wells tapping deep unconfined aquifers.
Water Resour. Res. 15, no. 5: 1167-1176.
165
APPENDIX E1
Solution Derivation
The Laplace-space expressions for the governing equations ([4a-b]) are
ph1 − γ 1 h0 = D1
d 2 h1
dz 2
(A1a)
d 2 h2
p h 2 − γ 2 h0 = D 2
dz 2
(A1b)
with the notation defined in the main text following equation (5).
The Laplace-space expressions for the boundary conditions are
h1 (− l , p ) =
hUB
p
dh2 (a, p )
=0
dz
for the constant-head condition (hUB) at the aquitard top
for the no-flow condition at the bottom of the aquifer
(A2)
(A3)
and
h1 (0, p ) = h2 (0, p )
(A4)
and
K1
dh1 (0, p )
dh (0, p )
= K2 2
dz
dz
(A5)
for continuity of head and flow, respectively, at the aquifer-aquitard interface.
The general solution to (A1) is
166
⎛ p ⎞
⎛ p ⎞ γ 1 h0
h1 ( z , p ) = A1 ( p ) cosh ⎜⎜
z ⎟⎟ + B1 ( p )sinh ⎜⎜
z ⎟⎟ +
p
D
D
1
1
⎝
⎠
⎝
⎠
(A6a)
⎛ p ⎞
⎛ p ⎞ γ 2 h0
h2 ( z, p ) = A2 ( p ) cosh⎜⎜
z ⎟⎟ + B2 ( p )sinh⎜⎜
z ⎟⎟ +
D
D
p
2 ⎠
2 ⎠
⎝
⎝
(A6b)
where Ai and Bi are functions of p that are determined from the boundary conditions.
Using the boundary conditions of (A2)-(A5) and the Di and Kr notation defined
after equation (5), expressions for A1, A2, B1, and B2 can be written as follows:
⎛ p ⎞
⎛ p ⎞
⎛ p ⎞
⎞
⎛ hUB
D2
⎜⎜
− γ 1 ⎟⎟ K r
l ⎟⎟ − (γ 1 − γ 2 ) tanh⎜⎜
l ⎟⎟ tanh⎜⎜
a ⎟⎟
sech⎜⎜
D
D
D
D
h0 ⎝ h0
1
1 ⎠
1 ⎠
2 ⎠
⎠
⎝
⎝
⎝
A1 ( p ) =
p
⎛ p ⎞
⎛ p ⎞
D2
l ⎟⎟ tanh⎜⎜
a ⎟⎟
Kr
+ tanh⎜⎜
D1
⎝ D1 ⎠
⎝ D2 ⎠
⎛ p ⎞
⎛ hUB
⎞
⎜⎜
− γ 1 ⎟⎟sech⎜⎜
l ⎟ + (γ 1 − γ 2 )
D1 ⎟⎠
h0 ⎝ h0
⎠
⎝
A2 ( p ) =
p
⎛ p ⎞
⎛ p ⎞
1 D1
1+
tanh⎜⎜
l ⎟⎟ tanh⎜⎜
a ⎟⎟
K r D2
D
D
1 ⎠
2 ⎠
⎝
⎝
B1 ( p ) = −
⎛ p ⎞ D1
A2 ( p )
tanh⎜⎜
a ⎟⎟
Kr
⎝ D2 ⎠ D2
(A7)
(A8)
(A9)
⎛ p ⎞
B2 ( p ) = − A2 ( p ) tanh ⎜⎜
a ⎟⎟
D
2
⎝
⎠
(A10)
Substituting (A7)-(A10) into (A6) and simplifying yields the Laplace-space
solution of equation (5a) in the main text.
167
Convolution Expression
Temporal variations in the head at the top of the aquitard (hUB) can be incorporated
using a standard convolution approach (e.g., Olsthoorn, 2008). In order to demonstrate
the approach for the head in the semi-confined aquifer (h2), equation (5c) can be rewritten
as
f 2 (z, p ) = F (z, p ) +
hUB
G (z, p )
h0
(A11a)
where
⎛ p ⎞
⎡ p
⎤
(z − a )⎥
cosh ⎢
γ 1sech⎜⎜
l ⎟⎟ + (γ 2 − γ 1 )
⎝ D1 ⎠
⎣ D2
⎦ +γ
F (z, p ) = −
2
⎛ p ⎞
⎛ p ⎞
⎛ p ⎞
1 D1
⎜
⎟
⎜
⎟
⎜
⎟
1+
tanh ⎜
l ⎟ tanh ⎜
a ⎟ cosh⎜
a⎟
K r D2
⎝ D2 ⎠
⎝ D1 ⎠
⎝ D2 ⎠
(A11b)
and
G (z, p ) =
1
1+
Kr
⎛ p ⎞
⎡ p
⎤
(z − a )⎥
l ⎟⎟
sech⎜⎜
cosh ⎢
⎝ D1 ⎠
⎣ D2
⎦
⎛ p ⎞
⎛ p ⎞
⎛ p ⎞
D1
l ⎟⎟ tanh⎜⎜
a ⎟⎟ cosh⎜⎜
a ⎟⎟
tanh⎜⎜
D2
⎝ D1 ⎠
⎝ D2 ⎠
⎝ D2 ⎠
(A11c)
The F function quantifies the dissipation of the pressure in the aquitard-aquifer system
produced by the barometric surface loading, while the G term quantifies the head change
produced by the boundary condition at the aquitard top. Only the G term is involved in
the convolution.
The infinite series expression for the convolution in real space is
168
⎡ N Vn ⎛ ln 2 ⎞ hUB (0 ) N Vn ⎛ ln 2 ⎞ ⎤
⎢∑ n F ⎜ z , k∆t n ⎟ + h ∑ n G ⎜ z , k∆t n ⎟ + ⎥
⎠
⎠ ⎥
⎝
⎝
n =1
n =1
0
A( z , t ) ≈ A( z , k∆t ) = 1 − ⎢ k −1
N
⎢
⎥
∆h
V ⎛
⎞
ln 2
n ⎟⎟
⎢ ∑ UBi ∑ n G ⎜⎜ z ,
⎥
⎣⎢ i =1 h0 n =1 n ⎝ (k − i )∆t ⎠
⎦⎥
where t=k∆t, hUB(0) is the head at the top of the aquitard at t=0,
∆hUBi = hUB (i∆t ) − hUB (i∆t − ∆t )
is the change in head at the aquitard top over one time interval ∆t.
169
(A12)
FIGURES
Well LEA5
12/18/07 to 1/23/08
Depth to water
Barometric Pressure
10
2.9
9.9
2.91
9.8
2.92
9.7
2.93
9.6
2.94
9.5
12/20 12/24 12/28
1/1
1/5
1/9
1/13
1/17
1/21
Date (tick mark at midnight)
Figure 1 – Depth to water from land surface and barometric pressure head for well LEA5
at the Larned Research Site for a period in the winter of 2007-08. Depth to water is
plotted increasing downward to display the inverse relationship between water level and
barometric pressure; spans of the left and right y-axes differ by a factor of ten.
170
Barometric Pressure Head (m of water)
Depth to Water (m below lsf)
2.89
a)
b)
Figure 2 – a) Location map and aerial photo of the Larned Research Site (LRS). Aerial
photo (year 2000) only shows wells discussed in paper, watercourse in photo is the
Arkansas River; b) High-resolution direct-push electrical conductivity (EC) log from near
the center of the LRS riparian zone. Wells in the High Plains aquifer are screened across
the interval marked A, while adjacent wells in the lower portion of the Arkansas River
alluvial aquifer are screened across the interval marked B. At this site, high EC values
indicate clays and low values indicate sands and gravels. Bar on right side shows the
vertical extent of the model discussed in the text.
171
High Plains Aquifer
Larned Research Site
3/12/03 to 3/12/08
3
Depth to water (m below lsf)
3.5
4
4.5
5
5.5
6
1/1/08
1/1/07
1/1/06
1/1/05
1/1/04
6.5
Date
Figure 3 – Depth to water versus time plot for LRS well LEC2 (see aerial photo in Figure
2a for location); well LEC2 has the most continuous record for this period of the three
wells shown in Figure 2a. The ovals indicate the time intervals used for the analyses
discussed in this paper. Note the pronounced seasonality of groundwater pumping in the
vicinity of the LRS and the period of significant recharge beginning in the latter half of
2006.
172
0.14
0.12
0.1
0.14
0.08
Barometric Response Function (-)
0.06
0.12
0.04
0.02
0.1
0
0
0.2
0.4
0.6
0.8
1
Time Lag (hr)
0.08
Well LEA5 - adjacent to east bank
Well LEC2 - 267.0 m east of LEA5
Well LWC2 - 419.5 m west of LEA5
0.06
0.04
0.02
0
0
0.2
0.4
0.6
0.8
1
Time Lag (d)
Figure 4 - One-day and one-hour (inset) barometric response functions (BRFs) for three
LRS wells in the High Plains aquifer in the winter of 2004. BRFs for winter 2005 and
2006 are similar in form. Agreement between the BRFs from these wells was observed in
all years since monitoring began (2001 or 2002). Error bars indicate one standard error
about the estimated functions; linear trend removed from data series prior to BRF
calculation.
173
0.26
Barometric Response Function (-)
0.24
0.22
Winter 2004
Winter 2008
0.2
0.18
0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
0
0.2
0.4
0.6
0.8
Time Lag (d)
Figure 5 - One-day barometric response functions for well LEA4 for winters 2004 and
2008; well LEA4 is adjacent to well LEA5 and screened across interval B of Figure 2b.
Linear trend removed from the data series prior to BRF calculation.
174
1
Winter 2004
Winter 2008
Theoretical Response Function
Kr Sensitivity
Barometric Response Function (-)
0.14
0.12
0.1
0.08
0.5 Kr
0.06
0.04
0.02
2 Kr
0
0
0.2
0.4
0.6
0.8
1
Time Lag (d)
Figure 6 - One-day barometric response functions for well LEA5 for winters 2004 and
2008, and the best-fit theoretical response function for the winter 2004 data. Hydraulic
parameters from the winter 2004 fit were used to generate the winter 2008 theoretical
response function. Estimated parameters were obtained from the first half-day of the
2004 BRF: Kr = 1.8 × 10-5 [-], D1 = 1.7 × 102 m2d-1, and γ1 = 0.97 [-]. The aquifer loading
efficiency (γ2), the aquifer diffusivity (D2), and the ratio of aquitard thickness to aquifer
thickness were fixed at 0.92 [-], 2.9 × 106 m2d-1, and 1.0, respectively. Similar results
were obtained for the other High Plains aquifer wells. The sensitivity of the response
functions to Kr is shown for variations of a factor of two about the 2004 theoretical
response function; similar variations in D1 produced plots that were barely
distinguishable from the response function, indicating the much smaller sensitivity to that
parameter for these conditions (i.e. aquifer and aquitard characteristics). Linear trend
removed from the data series prior to BRF calculation. Span of y-axis is half that of
Figure 5.
175